As memory makers prioritize shipments of their products to the AI sector, prices of solid-state drives for consumer applications increased significantly over the last few quarters, which naturally caused retail sales of SSDs to drop significantly in 2026. Nelson Duann, a vice president at Silicon Motion, one of the largest SSD controller makers in the world, said in an interview with Tom's Hardware at Computex 2026 that the retail drive market had almost disappeared in the first half of 2026. Nonetheless, the company still sold plenty of SSD controllers to module makers (the companies that make the finished SSDs), but those devices primarily went to PC OEMs.

"The retail SSD market has almost disappeared," Duann told us during the interview. "The controllers we sell to module makers are now largely ending up in SSDs that are shipped to PC OEMs. The reason is that OEMs cannot obtain enough NAND directly from memory manufacturers, so they are increasingly sourcing SSDs from module makers instead."

According to Duann, PC makers have to buy from SSD module makers because NAND vendors reduced allocation to the client/consumer PC market and redirected most NAND supply to data center products.

As a result, PC OEMs like Acer, Asus, Dell, and HP cannot get enough NAND or SSD supply directly from NAND manufacturers and have to turn to module makers for solid-state drives. The latter traditionally served end-users and had plenty of aftermarket products with enhanced performance and cooling, but now they increasingly serve PC makers instead.

"In the past, most module makers focused on the retail market," Duann said. "Since late last year and into this year, that has changed. OEM demand has become much stronger, and module makers are now supplying a significant portion of their production directly to PC manufacturers. As a result, most of the controllers we sell to module makers ultimately end up in systems built by PC OEMs."

The reallocation of NAND flash supply from the consumer segment to the AI data center segment has caused structural changes in the market in general, hurting some companies badly, but letting others benefit. For independent developers of SSD controllers — such as Phison and Silicon Motion — the situation is generally positive as demand for server-grade drives is increasing, whereas sales of client storage devices are not necessarily decreasing unit-wise. For more details, check out the whole interview with Nelson Duann on Tom's Hardware.

The storage market in general, and the SSD market in particular, has changed structurally in a matter of several quarters due to overwhelming demand for storage devices from the AI sector. As a result of the rapid price increase, demand for solid-state drives and other memory-based PC components dropped in retail in the first quarter. But as paradoxical as it might seem, shipments of SSD controllers increased in Q1 2026 so significantly that both Silicon Motion and its rival Phison posted record Q1 results. This happened not only because these companies increased sales of their data center and enterprise-grade controllers to hyperscalers and server makers, but also because they increased sales of all types of products, both to SSD producers and NAND manufacturers.

At Computex 2026, we sat down with Nelson Duann, Senior VP of Client & Automotive Storage Business at Silicon Motion, to find out how the company's business is going amid the ongoing structural market change and what to expect from SMI on the client storage front in the coming years.

We must say, Nelson Duann is a valuable speaker with an interesting background and a rather unique view on the market and technology development. He has been with SMI since 2007 and has led Silicon Motion's Client & Automotive Storage business since late 2023, when the company was reorganized into two business units and assigned new general managers for each. Duann is responsible for product strategy, OEM engagement, and program execution across these segments. Before assuming his current role, he headed marketing and R&D efforts for mobile storage products and played a significant role in expanding the company's SSD controller and mobile storage businesses into leading positions in their respective markets. Before Silicon Motion, Duann worked at Sun Microsystems on UltraSPARC processor development projects.

Without further ado, let's get to talking.

A rough time the industry, a good time for SMI

Anton Shilov: It has been a rough year for the industry in general, and memory-related segments of the industry in particular, so far. Yet, here we are, your first quarter sales were $342.1 million, up 23% quarter-over-quarter and 105% year-over-years, with sales of SSD controllers up 40% to 45% year-over-year. How did you manage to achieve that? At least on your side of the business.

Nelson Duann: There were two main drivers behind the growth of our business unit in the first quarter.

The first was our Ferri product line, where we sell complete storage solutions rather than just controllers. As NAND prices increased, the prices of those solutions rose as well. We performed particularly well in data center boot drives. On the automotive side, many NAND suppliers have made automotive a low-priority market despite strong demand from carmakers. Ferri is one of the few products that can still ensure supply, so we benefited both from higher NAND ASPs and from having inventory available to serve customers.

The second driver was our controller business. Sales of our high-end controllers, including PCIe Gen4 and PCIe Gen5 SSD controllers as well as UFS 3.1 and UFS 4.1 controllers, were very strong and contributed significantly to revenue growth.

We also started to see some contribution from our enterprise business. Volumes were still relatively small in the first quarter, but enterprise SSD controller programs are beginning to ramp and helped build momentum.

On the client side, demand for high-end controllers increased because NAND suppliers are allocating more NAND to data center products and less to the consumer market. As a result, PC makers have increasingly turned to module makers for SSD supply. While we do not sell SSDs ourselves, most module makers use our controllers, so as their shipments to PC OEMs increased, our controller shipments and revenue increased as well.

Anton Shilov: So memory makers are still supplying PC OEMs directly, but at lower volumes?

Nelson Duann: Yes. They are still supplying them, but allocations to the client market have been reduced significantly. Most NAND is now being directed to the data center segment.

Companies such as Lenovo, HP, Dell, Asus, and Acer cannot always obtain enough NAND from memory manufacturers. As a result, they increasingly turn to module makers for SSD supply. Those module makers build PCIe Gen4 and PCIe Gen5 SSDs, and because most of them use our controllers, that trend benefits our controller business.

Anton Shilov: So, NAND price increases have not really hurt your controller business; you have actually managed to sell more controllers than a year ago.

Nelson Duann: More than a year ago, and more than in the previous quarter, but primarily in the high-end segment. The low-end market has been affected because consumer demand is weaker. NAND has become a valuable resource, so PC makers prefer to allocate available supply to higher-priced products. As a result, demand for high-end SSDs has increased, and those products use controllers with higher ASPs. Growth in the high-end segment has more than compensated for weakness in the low-end market, which is why our controller revenue continues to grow.

Anton Shilov: So NAND shortages have actually helped you.

Nelson Duann: I would not say we are enjoying the shortages, but we have certainly benefited from them.

Anton Shilov: What do you expect to happen in the second half of the year?

Nelson Duann: I think supply will remain very tight. In fact, we met with NAND suppliers this morning, and they told us that the current shortage is only the beginning. They believe 2027 will be even worse.

For the second half of this year, I expect conditions to remain largely unchanged. Supply will stay constrained. You will hear analysts talk about declining PC demand and lower PC shipments, but demand itself has not disappeared. Apple continues to gain share in both PCs and smartphones, which shows that people are still buying devices. The problem is not demand; the problem is supply.

Many PC and smartphone makers cannot obtain enough components to ship the volumes they want. Because NAND and DRAM have become such valuable resources, suppliers are prioritizing higher-end products with higher ASPs and better margins.

I expect this situation to continue through the second half of the year. Looking into next year, NAND makers are quite pessimistic. They believe supply constraints will worsen because cloud service providers and data center operators continue to increase their demand. As a result, NAND suppliers are directing more of their output toward the data center market.

They still want to support consumer devices and automotive applications, but those allocations are relatively small and do not materially change the overall supply picture.

Anton Shilov: So the additional SSD controllers you sold in the first quarter were mostly sold to module makers rather than directly to OEMs.

Nelson Duann: Let me explain that in more detail. The retail SSD market has almost disappeared. The controllers we sell to module makers are now largely ending up in SSDs that are shipped to PC OEMs. The reason is that OEMs cannot obtain enough NAND directly from memory manufacturers, so they are increasingly sourcing SSDs from module makers instead.

In the past, most module makers focused on the retail market. Since late last year and into this year, that has changed. OEM demand has become much stronger, and module makers are now supplying a significant portion of their production directly to PC manufacturers.

As a result, most of the controllers we sell to module makers ultimately end up in systems built by PC OEMs.

Anton Shilov: One advantage Silicon Motion has is that you work with all six NAND manufacturers and also supply controllers to some of them. Given the current shortages, have shipments of controllers to NAND makers declined?

Nelson Duann: Unit shipments to NAND makers have declined. However, those companies are increasingly focused on higher-end products, so the mix has shifted toward premium controllers with higher ASPs. As a result, while controller volumes are lower, the higher ASPs have largely offset the decline in unit shipments. Consequently, our controller revenue has continued to grow.

Anton Shilov: One quick question. When you ship controllers to NAND makers, are they standard Silicon Motion products, or are they customized for specific SSDs?

Nelson Duann: NAND makers are no longer very active in the retail SSD market because NAND supply is limited. Most of the SSDs they build today are shipped to PC OEMs, and many of those drives use our controllers.

The controller hardware itself is generally not customized. What changes from customer to customer is the firmware. Every PC maker has different requirements, so NAND suppliers or SSD makers may ask us for firmware modifications tailored to a particular OEM. The controller ASIC remains the same; the customization is primarily done at the firmware level.

Anton Shilov: So the differentiation happens mostly through firmware tuning rather than hardware changes?

Nelson Duann: Correct.

Mobile Storage

Anton Shilov: Since you mentioned that the retail market has weakened considerably, can you provide a breakdown of your SSD, UFS, and eMMC controller business?

Nelson Duann: We group UFS and eMMC together and treat SSD controllers as a separate category. In the first quarter, the split was almost even. UFS and eMMC combined accounted for roughly 52%, while SSD controllers represented about 48%.

The reason is simple: smartphone volumes are still higher than notebook volumes. On a unit basis, mobile storage controllers ship in larger quantities than SSD controllers. We expect a similar mix going forward.

Anton Shilov: What is the split between UFS and eMMC today?

Nelson Duann: That is a good question. If we look at revenue, UFS generates more revenue than eMMC. However, in terms of unit shipments, eMMC volumes are still higher. The reason is simple: eMMC carries a much lower ASP. So while UFS contributes more revenue, eMMC still ships in larger quantities.

Anton Shilov: So eMMC still refuses to disappear.

Nelson Duann: Yes, that is true.

Anton Shilov: I assume that is because eMMC is used not only in smartphones, but also in a wide range of other devices.

Nelson Duann: Exactly. It’s widely used in IoT devices, automotive applications, and many other embedded products. That’s why eMMC will remain around for a long time. It has a very long tail.

Anton Shilov: Since your sales have been growing, can you comment on your market share in client SSD controllers?

Nelson Duann: In client SSD controllers, our worldwide market share is approaching 30% to 32%. For mobile storage, combining UFS and eMMC controllers, we estimate our share of the Android smartphone market at around 25% to 26%.

Anton Shilov: That is a significant share.

Allocation priorities

Nelson Duann: It is. And we expect it to grow further. NAND suppliers are allocating less memory to PCs and smartphones, which forces device makers to seek alternative sources of supply. Many turn to module makers, and we have a very strong position in that ecosystem.

As long as module makers continue to gain business, demand for our controllers should benefit as well, even though we do not manufacture SSDs or storage devices ourselves.

Anton Shilov: So the smartphone market is behaving somewhat like the PC market: device makers cannot get enough supply directly from NAND vendors and therefore have to turn to module makers?

Nelson Duann: Let me explain the allocation priorities in more detail. For NAND suppliers, the first priority is clearly the data center. The second priority is smartphones, because NAND is often supplied together with Low Power DRAM. The third priority is PCs, and the fourth is automotive.

Because of that priority order, smartphones are in a better supply position than PCs. However, supply is still not sufficient because the majority of NAND output is going to data centers. The data center segment may receive 70% to 80% of total allocation, leaving only 20% to 30% for smartphones, PCs, automotive, and everything else. As a result, even smartphone shipments could decline this year because of NAND and memory supply constraints, perhaps by 15% to 20% compared with last year.

Anton Shilov: Especially at the low end.

Nelson Duann: Yes. The low-end [smartphone] segment is largely gone, while the high-end segment continues to grow. PCs will probably decline more sharply in unit terms than smartphones because PCs face a more severe supply shortage.

There is one uncertainty, however: China is a different market. China has domestic NAND and DRAM makers, and their strategy is not the same as that of foreign memory suppliers. Because they receive government support, they also have a responsibility to help maintain the health of the local market.

Foreign suppliers generally follow the highest-return opportunities and can allocate most of their supply to data centers. Chinese suppliers cannot do that in the same way because the government can provide guidance and encourage them to support certain local industries.

Anton Shilov: So PC makers in China can get more stable NAND supply?

Nelson Duann: Yes, and DRAM as well. Most Android smartphone makers are now in China, and China is also home to Lenovo, one of the world's largest PC makers. Lenovo still has supply issues, but compared with HP, Dell, Asus, and Acer, its situation is less severe because it has support from local memory suppliers.

Anton Shilov: Is there any chance NAND supply will increase meaningfully in 2027 and help ease the shortage?

Nelson Duann: Bit supply will increase, but not enough to match demand. Demand could increase by around 100%, while bit growth will not come close to that level.

DRAM and NAND output will increase somewhat because of technology migrations rather than major increases in fab output. New fabs are being built, but they will not begin meaningful production until late next year or early 2028.

Suppliers can get some bit growth by moving to newer NAND generations, but demand is growing faster. As a result, the shortage will not improve next year. It will get worse.

Anton Shilov: Given high memory prices, undersupply, and limited ways to bring supply and demand into balance, what is the best strategy for a non-memory company like Silicon Motion on the client side? Do you adopt new 3D NAND types as quickly as possible, or slow client investments and focus more on data centers?

Nelson Duann: We have several strategies. First, we have to follow NAND vendors very closely and make sure our controllers support each new NAND generation quickly. Once NAND suppliers move to a new generation, they cut older generations quickly, so our controllers must be ready as soon as new NAND reaches the market.

Second, we need to encourage customers to move toward higher-end products. That is where the revenue and margins are, both for us and for our customers.

Third, as NAND becomes more expensive, customers may reduce capacity per device to manage cost. For example, a product that used to ship with 128GB may move to 64GB. But each device still needs a controller, so lower capacity per unit can increase controller demand.

Fourth, we need to continue advancing our technology. Nvidia's recent AI PC direction shows that the PC paradigm is changing. These systems need much more efficient data movement to feed their processors.

From my point of view, the market is moving beyond the traditional PC toward what we call a Personal Agentic AI machine. A system with 128GB of DRAM is not a conventional PC anymore. Today's PCs typically have 16GB or 32GB of DRAM.

To support this shift, storage architecture and performance need to improve significantly. That is why we introduced our new PCIe Gen5 DRAM-less controller with much higher random I/O performance. For AI workloads, when DRAM is not enough for KV cache, data needs to spill over to storage, which makes random I/O extremely important.

We are already re-engineering our architecture for this shift. Personal Agentic AI systems are expensive today, but costs should come down over time as the whole ecosystem redesigns storage and data-transfer architectures.

New NAND types support

Anton Shilov: Do you also expand the list of NAND types supported by existing controllers to address supply constraints?

Nelson Duann: Yes. We are known for supporting all major NAND types available on the market, including MLC, TLC, and QLC, and we will continue to expand support.

You never know which NAND type or supplier will have better availability. The more NAND types our controllers support, the more flexibility our customers have when dealing with supply shortages. We will continue to invest. In addition to supporting upcoming NAND generations, we are also looking at PCIe Gen6. PCIe Gen5 is the mainstream today, but the data center and enterprise markets are already moving toward Gen6.

Anton Shilov: The only platforms with PCIe Gen6 support right now are AMD's EPYC 'Venice' and Nvidia's Vera CPUs. Intel's Xeon does not support PCIe Gen6 yet.

Nelson Duann: Not yet. But in the data center, do not forget Nvidia. Nvidia is also pushing very hard for PCIe Gen6 because its [GPUs] are data-hungry monsters. PCIe Gen6 storage is not being driven only by AMD. Nvidia is also pushing the storage ecosystem to move to Gen6.

On the PC side, Gen6 is not here yet. PCs are still on Gen5, but we are preparing now. Controller development takes years, so we cannot wait until Gen6 CPUs arrive. That would be too late. We always start preparing several years in advance for new technologies and new interface chips.

Arm is a default choice for now, but RISC-V is being considered

Anton Shilov: With the SM2524XT, you support a 4.8 GT/s NAND interface.

Nelson Duann: Yes, on the NAND side. More NAND makers are moving to 4.8 GT/s interfaces, so that is important. Faster NAND also enables more efficient AI-oriented storage architectures, including KV-cache offload. Those workloads rely on faster interfaces and more efficient data transfer. That is what we are preparing for in the second half of this year and next year.

Anton Shilov: The SM2524XT has four CPU cores. Why did you move from three cores to four?

Nelson Duann: The previous SM2504XT had three cores. We moved to four cores to improve random I/O performance. Random I/O benefits from several things. First, a faster NAND interface. Second, SCA technology. Third, more controller compute power. With random I/O, addressing is random, so the controller must quickly translate logical addresses to physical NAND locations. That requires fast computation. The additional CPU core helps improve random I/O performance, which is important for KV-cache offload.

Anton Shilov: Do you continue to use Arm Cortex-R cores?

Nelson Duann: Arm remains a good partner, and as long as Arm continues to support us, we will continue using Arm cores. At the same time, we are internally studying RISC-V.

For storage applications, Arm and RISC-V can both provide enough performance. The bigger issue is the toolchain. If we move from Arm to RISC-V, our R&D team needs to change toolchains. Some customers buy our controllers but develop their own firmware, and they would also have to change toolchains.

So we are evaluating RISC-V carefully. Arm remains more widely used, and the ecosystem is mature. We have not formally adopted RISC-V in client products yet, but we are doing intensive internal evaluation.

Anton Shilov: What kind of performance improvement does the SM2524XT deliver in typical client workloads?

Nelson Duann: It depends on the benchmark, but in general, we are targeting roughly a 25% increase in random performance compared with the previous generation. Sequential performance also improves by about 25%. Faster NAND is part of the reason, and the additional core also helps. But just as importantly, performance per watt has improved. It is easy to increase performance by consuming more power. The hard part is improving performance while staying within the power limits of a client PCIe drive.

Our new generation delivers more performance per watt, perhaps around 10% to 15% better than the previous generation. That is important because we need to increase performance while keeping the entire client SSD within its PCIe power envelope.

Future SSD controllers

Anton Shilov: Is SCA coming to high-end eight-channel client SSD controllers?

Nelson Duann: For PCIe Gen5, no. Our first-generation Gen5 client controller was an eight-channel DRAM-based design. But our second- and third-generation Gen5 client controllers are four-channel DRAM-less designs. Since the NAND interface is faster, we no longer need eight channels to achieve the same or better performance.

For PCIe Gen6, however, we are considering going back to eight channels on the client side. For enterprise Gen6, the plan is 16 channels.

Anton Shilov: Future SSD controllers will have to support NAND with many more layers, which means stronger ECC will be required.

Nelson Duann: Correct. LDPC technology must advance generation by generation. We have a dedicated team working on error-correction capability because it is a fundamental part of NAND flash controller design.

As NAND moves to newer technologies and higher layer counts, we need stronger ECC to maintain reliability and data integrity.

Anton Shilov: Enterprise controllers have already moved to 16KB LDPC codewords. What about client controllers?

Nelson Duann: We are considering 16KB LDPC for our next-generation client chip, probably for PCIe Gen6, but we have not made a final decision yet. We are evaluating both the advantages and the disadvantages.

The advantage is that 16KB LDPC can correct more errors, which helps support future NAND. The downside is cost, die area, and compute requirements. Another issue is that most PC hosts still issue 4KB read and write commands, while 16KB LDPC is better aligned with larger data units. That mismatch can create read-modify-write operations, increase latency, and affect QoS. Enterprise hosts have moved much more toward 16KB commands, but client PCs remain mostly 4KB. So we need to evaluate the impact on performance, latency, and QoS before making a decision.

Anton Shilov: I assume different product lines also require different LDPC designs.

Nelson Duann: Exactly. Different product lines use different LDPC designs. For mobile devices, low power is extremely important, so LDPC is tuned for power efficiency rather than maximum performance. For PCs, there is a little more power budget, so we can tune LDPC more toward performance. Enterprise SSDs have even more power headroom, so performance becomes the top priority.

It is not one LDPC design for every product. We tune LDPC differently depending on the requirements of each market.

Anton Shilov: When do you expect to offer client PCIe Gen6 controllers?

Nelson Duann: Our current plan is for the end of next year. We are not pushing client Gen6 because of Intel or AMD CPUs. We are pushing it because of Nvidia. Nvidia is moving into the client side as well, and you can sense that from its keynote. Nvidia’s processors are power-hungry and data-hungry, so our client-side PCIe Gen6 roadmap is driven by Nvidia, not Intel or AMD.

Anton Shilov: Enterprise PCIe Gen6 controllers are coming out this year, correct?

Nelson Duann: Yes, the SM8466 is coming this year. Client Gen6 is planned for next year. We are not rushing to bring client Gen6 out this year.

Anton Shilov: The client Gen6 controller is called Neptune, right?

Nelson Duann: Yes, that is our codename.

PLC NAND may not be a viable option as QLC to reach 4Tb per device

Anton Shilov: Any thoughts on PLC NAND?

Nelson Duann: So far, there is no sign of activity around PLC. NAND makers are benefiting from the current market situation, so they are not eager to move to PLC.

QLC has already reached 2Tb per die, and we hear some companies are working on 4Tb dies without moving to PLC. If you move to PLC, the capacity of a single die becomes very large and more difficult to use.

PLC could reduce cost from the memory vendor's point of view, but cost reduction is not their main concern right now. In my personal opinion, memory vendors have learned from past cycles. Previously, when shortages occurred, they rushed to increase output, which later created oversupply and hurt everyone. This time, they are much more disciplined. If NAND makers wanted to move to PLC, they would need to tell us years in advance because we would have to build controllers that can reliably read that NAND. Right now, we see no sign of such a transition.

Anton Shilov: So, for ultra-high-capacity SSDs, the industry may focus more on advanced NAND packaging than PLC.

Nelson Duann: Exactly.

We learned a lot about Intel’s upcoming plans for desktop CPUs at Computex 2026. In classic Intel fashion, we’ve already heard a lot about the company’s next-gen CPUs, codenamed Nova Lake, even while the recent Arrow Lake Refresh CPUs are still warm from the oven. But on the ground in Taipei, we heard not only more about Nova Lake and the Z990 platform it’s arriving on, but also how Intel intends to handle the rollout and how it will fill the gaps in its lineup with “Raptor Lake Next,” which is supposedly slated to launch next year.

Trade shows are the best opportunity to learn details about unreleased products before they show up in a press deck, and simultaneously the worst venue to do so. With jet-lagged representatives and reporters, thousands of people whizzing past, and the threat of Jensen Huang showing up to sign components and shut down a floor on a moment’s notice, it’s easy for things to get lost in the shuffle. So, we’re going to work through everything we learned about Intel’s upcoming plans in stages, starting with details that are confirmed, and working toward more speculative murmurs.

Intel has a fairly aggressive consumer roadmap, which the company itself would tell you – and the company told us as much at Computex, as a matter of fact. Both Nish Neelalojanan, senior director of client product management, and the recently joined Alex Katouzian, executive VP and GM of client, played up Intel’s roadmap to Tom’s Hardware, and for good reason.

Chronologically, Intel’s plans look something like this: We’ll see the first Nova Lake SKUs roll out at CES 2027. A few months later, we’ll see a refresh on the LGA 1700 socket with “Raptor Lake Next” CPUs, and come Computex next year, Intel will launch a 52-core flagship Nova Lake SKU. None of that is confirmed by Intel, and we have varying degrees of confidence in each step of the roadmap, so take it as speculation for now. We’ll dig more into the details we have and what’s simply rumored below.

What about AMD?

Before Intel, we should at least look at why we’re not talking about AMD’s next-gen desktop plans. Basically, we don’t have a ton of information on Zen 6 CPUs yet, and even less information about Olympic Ridge, the desktop consumer lineup of Zen 6 chips. Computex didn’t change that fact.

At Computex, AMD revealed the Ryzen 7 7700X3D, relaunched the Ryzen 7 5800X3D, and brought the RX 9070 GRE to the rest of the world. Unlike previous years, AMD didn’t hold a keynote, where we might’ve seen a more concrete tease of Olympic Ridge; AMD has already teased Zen 6 broadly several times. Bigger Zen 6 news is likely at the company’s Advancing AI event next month.

Although AMD hasn’t said when Olympic Ridge will launch, we originally expected it in late 2026. Now, 2027 is very likely. AMD has shifted the Zen 6 conversation toward its EPYC Venice chips, and confirmed production ramp on Venice in May. Although AMD traditionally leads with a consumer launch at the turn of a new microarchitecture, it’s unlikely that Olympic Ridge will launch before Venice. Demand for CPUs is spiking in the data center for agentic AI workloads, after all, and AMD is adjusting accordingly.

Olympic Ridge probably isn’t top of mind right now, from both AMD itself and its partners. AMD laid the groundwork for a unified CPU architecture generations back, and Intel’s approach has been a bit more disparate across client and data center (although that’s been changing with releases like Xeon 6 and Xeon 6+). We don’t know when Olympic Ridge news will arrive, but it almost certainly follows far greater detail about Zen 6 in the context of Venice.

What’s confirmed

Let’s start with the concrete details about Intel’s future CPU plans. These are things we have direct evidence for, be it photos, our own hands-on time, or sources we’re extremely confident in. At least two Z990 motherboards were at Computex, a third is rumored, and we saw (and held) what looked to be a near-production model in a closed-door meeting. And from that, we can already tell a lot about Nova Lake.

First, the LGA 1954 socket, which has now been pictured (we were told not to take pictures, but someone else did the dirty work, it seems). It’s the same size as the LGA 1851 socket, measuring 45 mm x 37.5 mm, and it retains compatibility with existing coolers, which we were able to confirm at Computex. It features more pins, as the name reveals, and uses the 2L-ILM, or two-lever Independent Loading Mechanism. The picture of the socket circulating matches what we saw at Computex.

The motherboard we saw featured dual 8-pin EPS connectors, along with an 8-pin PCIe connector near the bottom of the board, which is said to provide auxiliary power to the CPU. We’ve seen a leaked photo of the Z990 PCH now, which is said to draw more power due to broader PCIe 5.0 support. The Z990 board we saw, at least, had three PCIe 5.0 M.2 slots, along with three PCIe 5.0 expansion slots. Short of perhaps specialized designs with extra M.2 slots, we expect Z990 to support PCIe 5.0 across the board.

As for the chips themselves, all that is confirmed from Z990 motherboards is that Nova Lake can scale up to a high-end power design. We’ll speculate more on specific numbers later, but we’ve seen auxiliary power beyond two 8-pin EPS connectors on two Z990 motherboards now, and the motherboard we held had an extremely high-end VRM design; we can’t say more than that at this point.

An important caveat here is that we’re dealing with high-end motherboards and discussing how high the platform can scale, not how it will scale. Plenty of ink has been spilled about Nova Lake’s supposedly high power draw, but we really don’t have details about the chips themselves, rather just the tippy-top of the platform that will support them.

Outside of Z990 boards, Intel has confirmed that Nova Lake is “coming at the end of 2026.” That’s what CEO Lip-Bu Tan said at the company’s full-year 2025 earnings call back in January. What we were told by multiple vendors at Computex is Q1 2027, with a portion of those vendors specifically pointing to CES 2027. Similarly, with Z990 motherboards, some vendors said Q1 2027 while others said Q4 2026 (one even hinted at Q3). Believe it or not, these timelines actually all match up.

What’s lost in translation here is when the sale is happening. Before Nova Lake launches publicly, Intel and motherboard vendors will need to sell products into the channel, which, a few months later, will be available for sale at retailers for you to buy. What we’re likely looking at is sales into the channel in Q4, a public launch of Nova Lake at CES 2027, and retail sales in Q1. When Tan says Nova Lake is coming at the end of 2026 to a group of investors, he’s likely referring to selling into the channel, not the final retail sale.

What’s likely

Now, we’re getting into a bit more speculation. These are some of the details we heard about at Computex, or confirmations of previous rumors that we don’t have any concrete evidence for. Given the conversations we had at Computex, and a healthy dose of critical thinking, these are the details that are likely but not confirmed. There’s always a chance we’re just blind men touching an elephant on some of these points.

First, Nova Lake. For nearly a year now, it’s been rumored that the highest-end Nova Lake SKU will scale up to 52 cores. That’s the number we heard at Computex, as well, but not as a typical flagship. Rather, we heard that Intel plans to lead Nova Lake with a 28-core flagship, which will launch at CES 2027, and introduce a high-end 52-core model later in the year. The timeframe we heard was Computex 2027, but if anything is subject to change, it’s a release date that’s a year away. For now, let’s call it later in 2027.

The 52-core SKU will apparently come with 16 Coyote Cove P-cores, 32 Arctic Wolf E-cores, and a cluster of 4 LP-E cores; we didn’t hear that at Computex, nor anything to the contrary, but that’s what has been previously rumored. That model will reportedly come with two compute tiles, so the 28-core model with a single compute tile will likely look like an 8 + 16 + 4 split. That’s pure extrapolation at this point, however.

As for the 52-core model, we were told it comes with a PL1 of 175W and a PL4 of up to 700W. The PL1 number is what’s important here. Although that is a sizable increase over the 125W PL1 of both the 285K and 14900K, 52-core Nova Lake doesn’t sound like a direct replacement for those parts. Given the timing and extra power demands, it looks more like a spiritual successor to Intel Extreme Edition chips, targeting enthusiasts with deep pockets and the HEDT crowd.

Nova Lake is treaded ground at this point, however. Something new we learned about from Computex is “Raptor Lake Next.” After hearing the name, we asked Intel, which declined to comment on Raptor Lake Next at this time. Apparently, however, it will be the third refresh of Raptor Lake CPUs on the LGA 1700 socket, particularly targeting budget-conscious builders while Nova Lake satiates the enthusiast crowd.

There are some pieces of circumstantial evidence that point to a reintroduction of LGA 1700 CPUs. First, this has been previously rumored. In April, prolific leaker Jaykihn hinted at another Raptor Lake refresh coming in 2027. We’ve now heard that the range is called Raptor Lake Next from multiple sources, and it’s specifically coming in the first half of 2027, some months after the initial Nova Lake launch.

Additionally, multiple motherboard vendors told us that they’re ramping production of LGA 1700 motherboards, including DDR4 boards, though they didn’t say it was in relation to any new CPU releases. Intel itself has dropped a few hints, as well. Earlier in the year, Intel’s Robert Hallock said that Raptor Lake will be “abundantly available” in the market, and at Computex, Intel’s Nish Neelalojanan told Tom’s Hardware that Intel “will continue to make sure that there are products which can take care of older memory technologies.”

It would certainly make sense for Intel to refresh Raptor Lake a third time. Although data center demand is offsetting it, the decline in desktop sales from high memory prices hits Intel and AMD on the balance sheet as well. Just about everyone we spoke with at Computex talked about the state of memory prices, and Intel has a DDR4 platform that it’s still actively selling on the market. AMD, with a hard switch to DDR5 with Zen 4, has to reach back further to revitalize DDR4 options, but Intel already has a small ecosystem of DDR4 motherboards and CPUs available now, which it could easily bolster. We’ve heard that bolster is coming in the opening months of next year.

What that range looks like remains a mystery, however. It could be a proper refresh, or it could simply be an infusion of 14th-gen stock (and LGA 1700 motherboards) into the market along with new price points; both Raptor Lake generations have slowly crept up in price since the end of last year. The important thing here is that it seems Intel is targeting LGA 1700 for the lower end of the market, as Arrow Lake, with its underperformance and high price due to exclusively using DDR5, won’t provide the last-gen value bridge that previous generations have.

After Tom's Hardware originally broke the news about Raptor Lake Next, we followed up with Jaykihn, who provided a few specs.

*Naming unconfirmed by Intel, specifications rumored

The specs we've heard about are for the four SKUs above, which would comprise the main lineup of chips with integrated graphics enabled; apparently, Raptor Lake Next will include options with the iGPU disabled, as well as mobile chips. The final branding is unconfirmed, but we've heard that Intel intends to launch under the Core Ultra 200 name.

Out of the four SKUs, the 16-core Core 5 looks like Intel's breadwinner. Throughout 12th- to 14th-Gen, Intel topped out Core i5 models at 6 P-cores. You'd have to step up to a Core i7 for 8 P-cores. If these specs are correct, Intel is stepping down to an 8 P-core configuration a tier in branding, which will hopefully come with a cut to price.

What’s still up in the air

Some of the finer details of Nova Lake are still up in the air. That is, we don’t have any direct evidence for them, nor any corroboration from Computex. That’s not to say that the details here are false. Rather, we just need more information to say, for sure, that some of these details are a part of the Nova Lake lineup.

First and most obvious is bLLC, or big Last Level Cache. This is one of the earliest Nova Lake rumors that is still circulating, and for good reason. Intel hasn’t found an effective counter to AMD’s 3D V-Cache CPUs in more than four years. We’re closing in on half a decade where AMD has entirely owned the high-end of PC gaming, which has continually eaten away at Intel’s market share. bLCC is, apparently, Intel’s counter to 3D V-Cache, using its own Foveros 3D hybrid bonding to stack additional last-level cache.

Tom’s Hardware asked Intel CEO Lip-Bu Tan and a panel of executives at the company how it plans to address X3D CPUs, and Alex Katouzian, a 20-year Qualcomm veteran who recently joined Intel in a leadership role over the client group, said the following: “When I first came in and started reviewing road maps for the team, I was very pleasantly surprised. So, stay tuned, a very strong roadmap [is] coming, and we will be gunning for that section of the market as well. And so, please stay tuned.”

Context is important, but Katouzian is really only saying that Intel is gunning for high-end gamers with its roadmap, which, of course, it is. Otherwise, bLLC has entirely been a topic of the rumor mill. Intel has indirectly teased it with PR hits about its packaging capabilities, but that extends far beyond bLLC. Hybrid bonding, especially from a foundry perspective, has far greater legs in the data center.

Although Intel has the packaging and bonding capabilities, the scale of them for a mass-market product like Nova Lake is questionable. Intel would need to bond the SRAM to the logic tile with Forveros and package the chip with EMIB, creating the “EMIB 3.5D” combination that Intel has talked about previously. We first saw EMIB 3.5D on the Ponte Vecchio data center GPU, but most recently and relevantly on Clearwater Forest, Intel’s first foray into putting 18A in the data center. The capability is there, but if Intel can scale that up to a consumer range with more limited die space and higher per-core performance remains to be seen.

One advantage of Intel’s hybrid bonding and advanced packaging is that it can package dies from other foundries, not just those from Intel foundries. That brings us to the second finer point about Nova Lake, which is the node. Originally, the assumption was that Intel would use 18A for Nova Lake. We have 18A on mobile with Panther Lake, in the data center with Xeon 6+, but not on the desktop. Further, Intel has previously commented about reshoring its manufacturing for consumer chips after a brief stint with TSMC for logic tiles in both Lunar Lake and Arrow Lake.

Around this point last year, however, rumors started circulating that Intel is using TSMC’s N2 for Nova Lake. The source of the rumor is flimsy, however. Well-known reporter Charlie Demerjian of SemiAccurate reported in July 2025 that Intel taped out a major product. The report didn’t mention what product, what foundry, or even include “TSMC” anywhere on the page. Still, other outlets took the story, claiming that not only was Demerjian talking about Nova Lake, but also that he was talking about TSMC N2.

There are reasons Intel could use TSMC for the logic die. The company has reiterated that it’s shifting wafer capacity toward the data center, so if TSMC can fill additional capacity on the desktop, we could see TSMC on the main logic die. It’s also possible that TSMC is manufacturing other tiles on Nova Lake. Intel has consistently blended nodes in recent generations, so even if Intel were to confirm that it’s tapping TSMC for Nova Lake, that doesn’t necessarily mean the Taiwanese giant is manufacturing logic.

And, just as easily, Intel could absolutely be using TSMC for logic. That’s the point here; we really don’t know at this point, outside of vague reporting, getting swept up in the rumor mill, and taking on a life of its own. The Cinderella story for Intel would be Nova Lake on 18A, but given the struggles on 18A yields, it wouldn’t be surprising to see TSMC at the helm for Nova Lake once again.

Hurry up and wait

Intel needs a much more aggressive roadmap on the desktop than AMD, frankly, and that roadmap is starting to take shape. Although AMD and Intel compete on the finer points of performance, Team Red has almost exclusively taken market share away from Intel, quarter over quarter, for the past decade. There are only a handful of quarters in that time when AMD has lost market share, which it has always rebounded from in the quarter that follows.

Even if Intel still represents the majority of the desktop market — and it does based on the latest market research — the trend is abundantly clear. Add on top of that clear fumbles like Arrow Lake, and it’s obvious that AMD doesn’t need to move the needle much to continue swiping customers. Intel needs to make big moves to recover.

We should have more official details about those plans soon. Intel mostly sat Computex out on the consumer front, short of the Arc G3 range that, although exciting for gaming handhelds, is destined to be a niche product given the high prices of the devices those chips are going in.

For the past four years, Intel has held its Tech Tour event in the fall, taking the place of its previous Architecture Day, which took place in the late summer (most of those details have shifted to the Hot Chips conference in August). Intel has already told us that Hot Chips will have more details about Diamond Rapids, Intel’s next-gen P-core Xeons. That leaves Tech Tour for when we’ll likely get a full architectural deep dive on Nova Lake. Intel has yet to confirm Tech Tour 2026, but we have no reason to believe the company will sit out the rest of the year at this point. It also lines up with what we’re hearing about Nova Lake’s release — architectural details in the fall, a launch at CES 2027, and availability in Q1.

Regardless of when the exact dates fall, Computex made it clear that Intel is readying Nova Lake for a release soon. Multiple motherboard vendors brought Z990 motherboards to Computex and actively showed them to the press; I can’t imagine that was sanctioned by Intel.

As for Raptor Lake Next, Computex is the first quasi-confirmation we’ve heard of the range. That name apparently appears on Intel’s roadmap at some point in the first half of next year. With Nova Lake at the high-end and Raptor Lake Next in the midrange, Intel might have a one-two punch strategy to earn back some spots in the market, especially as AMD turns its Zen 6 focus toward the data center and prioritizes older architectures on desktop, given high DDR5 prices. Now, we just need to wait and see how those internal plans materialize as the rest of the year goes on.

We managed to grab pics of the newest upcoming server sockets from both AMD and Intel at Computex 2026. Both AMD and Intel are preparing to launch their next-generation server platforms that use all-new sockets, which enable new levels of performance, functionality, and power delivery.

AMD is a bit ahead with its SP7 platform in 2026, while Intel’s gargantuan 9324-pin socket will be used for Xeon ‘Diamond Rapids’ in 2027. While the platforms are entirely different, what makes them similar is the massive dimensions of CPU sockets and coolers.

AMD’s SP7 is the company’s next-generation socket that will support AMD’s 6^th Generation EPYC ‘Venice’ processors with up to 256 cores. The socket is huge and is rumored to support 16 DDR6 memory channels using 12.8 GT/s MRDIMMs as well as up to 96 PCIe 6.0 lanes (with the CXL protocol on top, though this is a processor, not a socket feature).

Based on information from Auras, the SP7 socket will be able to handle CPUs with a peak power consumption of up to 1,400W, so Auras and other companies are prepping liquid cooling solutions for these parts. In person, the socket is strikingly large, occupying most of my palm and overshadowing today’s server CPU packages.

Given the fact that the socket must support so many memory channels and PCIe lanes, it is not surprising that it is that large. Despite its enormous dimensions, the socket is still compact enough to enable dual-socket server designs, so AMD’s partners will be able to offer systems with up to 512 x86 cores as soon as its next-generation EPYC processors arrive later this year.

Meanwhile, for those systems that do not need so many cores and memory channels, AMD is prepping the SP8 platform that is set to offer fewer cores and DDR5 channels. Interestingly, Auras is working on water blocks for SP8 sockets as well, which means that the platform will still be quite mighty in terms of power consumption.

But while AMD’s SP7 is huge, Intel’s 9324-pin socket easily dwarfs it, as it is noticeably longer than the palm of my hand. The socket will work with Intel’s Xeon ‘Diamond Rapids’ processors with up to 192 cores, a 16-channel DDR5 memory subsystem supporting MRDIMMs, and PCIe Gen6 lanes.

Intel is yet to announce the processor base power of Diamond Rapids processors, though, since Auras is prepping water blocks for these CPUs, we're talking about circa 300W – 500W PBP and over 1 kW peak power consumption. Meanwhile, given that the socket is so massive, we would not be surprised if Intel’s 9324-pin socket will also support the Coral Rapids processors, presumably due in 2028 – 2029.

While hyperscalers rush toward expansion amid the swelling demand for AI data centers, Marvell last week shared its vision for an optical interconnect solution that can theoretically pool resources between discrete data centers across thousands of kilometers.

Optical interconnections are steadily being deployed across the industry, over both short and long-distance connections, and we're going to be seeing much more in the future, according to Matt Murphy, Chief Executive at Marvell, speaking at Computex 2026.

"Imagine future data centers, a globally optically interconnected data infrastructure," Murphy said. "These rigid boundaries we have today, and the systems we have, they begin to disappear. Compute can now be pooled, memory can be pooled, and infrastructure can be composed dynamically at scale."

Constrained by distance

Murphy says that workloads no longer fit within one data center, which is why hyperscale cloud service providers increasingly need to build entire campuses consisting of multiple data centers connected by high-speed links, as clusters are becoming larger than a single data center.

Today, connecting multiple data centers within a single campus is not easy or cheap, but relatively straightforward. However, Marvell envisions that in the future it will need to connect data centers that are located at considerable distances from one another.

This is why Marvell is working on coherent optics and long-haul scale across optical networking technologies, which will connect data centers separated by thousands of kilometers. Marvell already has products which enable such connectivity today, including the Colorz 1600 1.6 Tb/s coherent optical solution based on a 2nm DSP, which targets inter-data-center connectivity and will sample later this year.

In addition, Marvell says it will offer the Ara 1.6 Tb/s family of interconnect solutions for data centers (with 3nm DSPs) as well as the Teralynx T100 102.4 Tb/s Ethernet switch, which supports 512 ports running at 200 Gb/s or 64 ports running at 1.6 Tb/s.

Murphy argues that today's architectures are constrained by distance because of copper interconnects: CPUs sit near memory because latency matters, GPUs sit near memory because bandwidth matters. As a result, workloads must be partitioned according to those physical limits. The head of Marvell claims that once optical interconnects penetrate scale-up interconnects, scale-up domains will not be limited by copper cable lengths, and those constraints will begin to disappear.

Nowadays, scale-up AI solutions, such as Nvidia's NVL72, are connected using copper wires, but scale-out connections tend to use optical interconnects. Once the number of AI accelerators within scale-up systems increases, they will also have to move to optical links, according to Marvell. This means that virtually all data center-grade interconnections will become optical, which might inspire hardware developers to reconsider the architecture of data centers.

Pooling resources

Murphy presented a rather interesting vision: firstly, optics will expand scale-up domains from 72 or 144 accelerators to 1,000 or more. But after that, optical connectivity will enter servers themselves. This will enable developers to disaggregate CPUs, accelerators (Marvell calls them XPUs), and memory into separate pools as distance will no longer matter, enabling better configurability and utilization.

"It is a data center without distance, where compute, memory, networking, and photonics operate as one unified system, where millions of resources across the data center can work together as if they were one machine," the head of Marvell said.

Keeping in mind that hyperscalers deploy hardware worth billions of dollars, even a 10% higher utilization will save a lot of money, and companies like Nvidia are clearly paying attention.

"In today's systems, the ratio of CPU and XPU or GPU is fixed, so these ratios have to be defined at the time the system is built and deployed, but no two workloads require exactly the same ratio," Murphy stressed. "Imagine a completely disaggregated architecture, XPUs in one system, memory in another, generic CPUs in another."

Today, companies buy something like an NVL72 system and get a fixed ratio of CPUs, GPUs, and memory, which may be efficient for certain workloads and inefficient for others. In the future, operators will be able to assemble a virtual machine from shared pools of systems, allowing for customization and flexibility, based on the type of workload. If a workload needs more memory than compute, operators often have to buy additional GPUs just to get the extra HBM, but they may just get memory in the future if Marvell's vision comes to pass.

"Once we decompose the system into separate pools of compute, memory, and they are all optically interconnected, we can then compose dedicated systems on the fly, which are then optimized wherever the workload is," Murphy said. "For the first time, architects can begin designing AI systems around the needs of the model, not around the limits of the interconnect."

One detail

While Marvell has the know-how to interconnect data centers across thousands of kilometers and technologies that enable pooled data centers, these visions do not necessarily intersect. Data centers located thousands of kilometers away cannot share resources — a 1,000 km round-trip takes light 10ms — which makes such long-distance resource sharing inefficient from a latency point of view.

However, Marvell's technologies enable hyperscale CSPs to synchronize AI campuses, access distributed storage, replicate data, and perform other operations that do not depend on latency. Meanwhile, the synchronization of AI campuses on different continents in a matter of hours could be a killer app for hyperscalers.

Some of the most interesting topics at Computex happen around the announcements themselves, and that’s certainly the case when it comes to Intel’s plans beyond Nova Lake. At the event, Tom’s Hardware heard that Intel has plans to launch “Raptor Lake Next” CPUs within the first half of 2027. The CPUs would mark the third refresh of Intel’s Raptor Lake range, which first debuted with 13th-Gen chips. The chips will reportedly live alongside Intel’s upcoming Nova Lake range, which the company intends to introduce at CES next year.

We don’t know the proper name of the CPUs yet, just that they’re referred to as “Raptor Lake Next” and will arrive some months after the first Nova Lake chips have hit the market. At least two motherboard vendors confirmed to Tom’s Hardware plans to increase production of DDR4 motherboards, both for AM4 and LGA 1700, citing increased DDR4 demand. They did not comment on Raptor Lake Next.

At this time, we simply know the name and a rough timeline when the chips are set to release, which follows a rumor that circulated in April about Intel’s plans to re-introduce Raptor Lake CPUs, as well as comments from Intel’s Robert Hallock that Raptor Lake will be “abundantly available” in the market. At the moment, Raptor Lake Refresh still holds the distinction of being Intel’s best CPU for gaming, with the new Core Ultra 7 270K Plus just narrowly falling short of the Core i9-14900K in games.

The specifications remain a mystery, as well as what exactly the range will look like. Recently, Intel introduced Bartlett Lake for embedded and industrial applications, which use exclusively P-cores and slot into the LGA 1700 socket. The flagship Core 9 273PQE goes up to 12 P-cores, four more than the Core i9-14900K. Bartlett Lake chips are socket-compatible with Raptor Lake platforms, though not supported through software.

Nonetheless, some enthusiasts have managed to get Bartlett Lake chips working on consumer 600-series and 700-series motherboards. Although we don’t know if Bartlett Lake will make an appearance for consumer applications under a different name, the mere existence of the range confirms Intel continues to produce Raptor Cove-based wafers on Intel 7.

We’ve corroborated the name Raptor Lake Next, but we still don’t know if it will be an entirely new range of processors. AMD recently reintroduced the Ryzen 7 5800X3D, turning back to DDR4 amid memory shortages, and it makes sense for Intel to do the same. That could simply look like an infusion of stock into the market and new price points, however.

Keep in mind that plans can change. Although we’ve heard the name from multiple sources, as well as confirmed an LGA 1700 ramp with vendors, even small details can change weeks before launch. If, say, memory prices drop severely in the next few months, that would almost certainly change Intel’s plans. For now, however, this is the rollout we’ve heard about.

Intel declined to comment on Raptor Lake Next at this time.

In the spirit of the times, Biwin, a well-known producer of solid-state drives and memory modules, announced that it had signed a two-year 3D NAND memory agreement worth $1.86 billion, the company told us at Computex 2026. We then dug up the announcement in a filing with the Shanghai Stock Exchange. The amount of money the company plans to spend on flash memory exceeds 50% of Biwin's annual revenue, demonstrating how committed SSD makers are to securing NAND supply amid shortages. Biwin's disclosure comes after several other companies assumed massive amounts of debt to secure supply, and crushing shortages have forced customers into ever-longer long-term supply agreements (LTAs).

Under the terms of the agreement, Biwin will purchase $1.86 billion worth of 3D NAND memory from an unknown supplier over a period of 24 months that starts on June 30, 2026. Both bit volume and pricing are fixed, so Biwin will get its memory at the fixed price no matter how high spot or contract prices are at the time. Of course, this represents risks if NAND prices drop in the next two years, though industry observers believe that NAND supply to SSD module makers will worsen in 2027, so Biwin's risks seem to be manageable.

"The total committed purchase amount under the Contract is US$1.8608 billion, and the commitment period is 24 months," Biwin's filing with the Shanghai Stock Exchange reads. "In accordance with the Contract, both quantity and price are fixed. The Company will complete purchases in batches from the third quarter of 2026 through the second quarter of 2028."

Biwin is among the world's biggest branded suppliers of SSDs, though with a 10% market share in 2024, it cannot rival Kingston or NAND makers themselves. The company's willing to spend $1.86 billion could enable it to grow its market share, or just supply SSDs to large customers, such as hyperscalers that demand a stable supply. At the same time, the purchase commitment that represents over 50% of Biwin's annual revenue shows how difficult it is becoming for the company to get its NAND, with some even commenting that the spot market could dry up in the future as companies are forced to sign LTAs instead. The big question is whether other SSD makers without their own NAND memory will follow through.

What strikes the eye in the announcement is that the procurement volume for 2026 accounts for 4.45% of Biwin's total NAND flash purchases in 2025, which indicates that the company is paying a fortune for a relatively low volume of NAND. Meanwhile, the procurement volume for 2027 accounts for 14.88% of the company’s total NAND flash purchases in 2025, which further underlines the increased pricing of NAND.

It should be noted that for $1.86 billion, the company is 'locking in a portion of its baseline demand for the next 24 months,' which means that the amount of NAND memory it gets does not cover all of its needs. Yet, there is good news too: give the large scale of the company and its roadmap, the overall risk of the supply agreement is 'considered manageable.'

So many PC cooling companies copy each other’s designs and features – I must have seen a dozen AIOs at Computex with curved screens and / or unified fans. But at Levelplay’s event last week, I was struck by a couple of concepts the company is working on. First was the MagBracket fan design, where the RGB lights are housed in a mounting plate, which connects via a USB-C cable that unifies PWM and RGB connections.

This allows the fans themselves to connect magnetically to the plate, and electrically via pogo pins. Drop them on the powered plate, and the fan lights up as it begins to spin. That would make cleaning your case fans exceptionally easy, and it also means the spinners are reversible, with connectors on both sides. So you can pop a fan off, flip it over, and put it back on to go from intake to exhaust.

That also means you can install the brackets on their own, and pop the fans on after your system is more or less built. Anyone who’s ever put a fan on a radiator or case in the wrong orientation (that has to be everyone who’s built more than one PC, right?) can rejoice at how easy that situation would be to fix here. Grab the offending fan, flip it over, pop it back on, and you’re good to you. You wouldn’t even need to reboot.

We don’t know pricing or availability yet, and the orange crossbars probably aren’t doing the airflow any favors. But I love the simple modularity of this design.

Next up, the Newtro 360 is a retro-styled cooler that puts a big tactile knob on top of an AIO, letting you adjust the fan speed, with a max cooling rating of 350W.

It’s certainly a more unique take than adding a screen or a VRM fan on top of the cold plate, but I’m personally not certain how often I’d be willing to pop my side panel off to tweak my AIO’s performance. I do like the orange accents on the sleeved cables and the old-school metal fan grilles, though.

Like the magnetic fans, the Newtro AIO is a concept, so we don’t know if or when the company will release the cooler, but it’s certainly the most tactile AIO I saw at Computex 2026.

Lexar's Regional Manager for Australia & New Zealand, Chris Xia, said that he personally thinks RAM prices are expected to double by the end of the year. While there have been a few instances of RAM module prices stabilizing and even some retailers selling them for a discount or bundled with other PC parts, he said that this is just sellers making an effort to make their inventory move and make way for new stocks, which are expected to be priced higher.

The current AI build-out is siphoning all the memory chips available from the traditional big three suppliers — Samsung, SK hynix, and Micron — with nearly all production capacity getting allocated towards high-bandwidth memory. Consumers are getting left behind, and as supplies dry up, their prices continue to go up.

Sapphire PR manager Edward Crisler said last December 2025 that DRAM prices will begin to stabilize in the next six to eight months, albeit at higher prices. However, the situation seems to have changed, and industry insiders now expect prices to continue going up towards the end of the year. Xia said that industry costs take up to eight to nine months to affect consumer prices, and the current market pricing trend has only been going up.

Some consumers get hope when they see RAM kits getting discounts or retailers lowering the list prices of these items, but Chris said that these are often the result of sellers trying to get rid of old inventory. They do this so that they can get some liquidity back and to make way for new stocks coming in from suppliers, usually at a higher price. Another thing that adds to the confusion is that some distributors manage to get their hands on unsold inventory from other regions that are still priced lower compared to what’s arriving now. Because of this, they’re able to sell at a lower price — but only until supplies last. Once the old stock runs out, they will eventually be forced to increase retail prices as market forces catch up with the low supply and high demand. Xia recommends that if you need to buy RAM, you should buy it now. Don’t wait for lower prices as they won’t arrive for years to come.

The memory chip crisis is going beyond desktop computers and laptops, which are expected to see shipments contract by more than 10%. Motherboard sales have already collapsed by more than 25% as the increasing RAM and SSD prices are making enthusiasts think twice before building a new system. Smartphones are expected to either get more expensive or see lower and slower memory configurations, and even action camera manufacturer GoPro is in trouble due to memory chip shortages and lower sales.

A new PC case brand is expected to arrive in North America, and it’s already gearing up with a couple of new and interesting designs. FormulaV Line, which quietly launched in 2024, announced two new cases at Computex 2026. This includes the Air Power G10, which includes three tilting fans that lets you direct the airflow inside the case where you need it, and the Crystal Z3, with its bottom-intake chamber that allows the case to easily breathe from four sides while still giving you a panoramic view of your build.

The Air Power G10 is FormulaV Line flagship case at Computex 2026, and its biggest innovation is the addition of three tilting front fans that allow you to change their angles by up to 10 degrees up or down. This would allow you to focus the airflow exactly where it’s needed — if you run an AIO cooler for your CPU, you could focus all the front fans towards your GPU to ensure that it gets the maximum amount of fresh air from outside your case. But if you prefer traditional air cooling for your processor, you can then point one of the fans towards it to help it get its fair share of cooler outside air.

This mid-tower case can accommodate ATX motherboards and comes with three 120mm front fans and a single 120mm rear fan. You can also install three 120mm fans at the top, up to two 140mm fans on the side, and two 120mm on the bottom. The top slot also accommodates radiators from 240mm to 360mm. Aside from the usual white and black cases, the Air Power G10 is also available in several other colors, including silver, dark blue, rose gold, and black wood.

Launching alongside the G10 is the Crystal Z3, which features a panoramic design. This PC case has two bottom-mounted fans that intake air from an open chamber, ensuring that it will never be starved for air, no matter what surface you plant your desktop computer on. Now, because the chamber is quite small, it would likely be difficult to reach inside for cleaning. So, FormulaV Line made the two bottom fans easily detachable, making maintenance and cleaning so much easier.

The G10 is expected to become available starting in September, while the Z3 will drop by August 2026. Unfortunately, we don’t have exact pricing details for these PC cases yet, but the company is known for its affordable pricing. The Air Power G10 is estimated to cost around $150, while the Z3 is projected to be around $80, giving budget PC builders a panoramic case option that won’t break the bank. These two new cases offer gamers more options, especially for those who can’t find anything they like in our list of the best PC cases.

A new PSU, cooling solutions, and a gaming chair

Aside from these two flagship PC cases, the company also has several other products on display at its booth. These include the FV1000GM Elite, a 1000-watt PSU with 80 Plus Gold efficiency certification, a fully modular design, and is compatible with ATX 3.1 and PCIe 5.1. It also boasts a Hybrid Fan Mode, allowing you to switch between maximum cooling and a quieter mode that adjusts the fan speed based on the temperature and other factors. This PSU is also expected to arrive by September 2026, and we’d love to get our hands on it to see how it compares against the best power supplies.

FormulaV Line also showed off a plethora of cooling solutions at Computex 2026. There’s the Ice Jet Vision 360 — an AIO solution with a 3.95-inch screen right on the CPU block — and the Air Bridge VT Vision, a three-fan solution that comes together as one unit and comes with three 3.38-inch TFT LCD screens right on the side, making it a great way to customize your PC case anyway you like. There are also several Ice Nova air coolers for those who prefer simpler setups, and they’re available in various sizes to ensure that you have an option no matter the size of the CPU case you’re building in.

Last, but not least, we saw the Solen gaming chair — an electric couch on wheels that lets you adjust its recline and footrest angle using rocker switches for the ultimate convenience. It also has a built-in vibrate function to soothe your sore back after a long day of gaming, and even has a USB-C and USB-A port for charging your gadgets right on it. The chair is powered by a removable battery, so you don’t have to worry about being tethered to the wall just to use all these functions. You can also top it up via USB-C, meaning you can keep it powered using your phone charger.

AMD's EXPO Ultra Low Latency program, announced at Computex 2026, aims to give users a one-click route to lower memory latencies than its existing EXPO profiles, but the company's initial announcement was light on details. To learn more about EXPO ULL, I stopped by G.Skill's Computex booth, where the company demonstrated four new kits that offer EXPO ULL support.

Memory latency directly affects how long the CPU has to wait in order to get data back from RAM, and so it has a major impact on CPU performance. But even as new DDR standards and ever-faster DIMMs have boosted memory bandwidth, DDR latency has improved at a much slower pace over time.

For some very high-level background, when selecting memory, PC builders will generally consider a given memory kit's speed and its CAS latency (CL). If you compare two CL30 memory kits, for example, the one with the higher clock rate will have a lower effective latency in nanoseconds (because CL30 expresses a number of clock cycles).

Knowing this, your first instinct for reducing latency might be to seek the highest-clocked memory you can find with the lowest CAS latency (like DDR5-8400 or even faster modules).

But on modern AMD platforms, it's not that simple. Reaching memory speeds higher than 6000 MT/s generally requires the use of a 1:2 multiplier mode between the clock of the integrated memory controller (the UCLK), which generally tops out around 3000 MHz, and the memory clock (MCLK). This 1:2 multiplier adds latency, and so it can counterintuitively reduce performance even as memory speeds climb above 6000 MT/s. (Remember that DDR memory moves bits at twice the clock rate, hence MT/s).

With this 1:2 multiplier active, by the time additional memory clock speed even begins to bring latency back down to where it would generally be in the 1:1 mode, you're looking at wildly expensive and exotic memory kits, and so most enthusiasts running Ryzen 7000 and Ryzen 9000 CPUs consider it desirable to choose memory that lets them run the UCLK and MCLK in 1:1 lockstep for the best balance of low latency and (relatively) low cost.

All that is why using memory faster than 6000 MT/s on AMD platforms is generally counterproductive for gaming performance. That's why modules in the range of DDR5-6000 CL30 are widely regarded as the overclocking "sweet spot" for Ryzen 7000 and Ryzen 9000 CPUs.

But that doesn't mean there isn't further room for improvement, as the introduction of EXPO ULL suggests.

G.Skill told me that until now, DRAM module makers were only permitted to change the four primary timings within EXPO (and XMP) profiles, leaving performance on the table. EXPO ULL affords memory makers more freedom to adjust the sub-timings within each of those four primary timings for even lower latencies, and to include those results in the memory's SPD.

Tweaking memory sub-timings on Ryzen platforms using community-made tools to determine the best potential settings used to be a fairly common practice for those seeking the best performance from their AMD systems, but Ryzen X3D processors and their massive slices of 3D V-Cache reduced those CPUs' sensitivity to those finer adjustments. It's become much more common to just get a DDR5-6000 CL30 kit, enable EXPO, and call it good.

But if you are focused on achieving the absolute lowest memory latency, EXPO ULL removes the need to perform the (tedious and tricky) process of determining those improved sub-timings by allowing memory makers to shoulder that work and include it as part of the one-click boost that EXPO provides.

For all that, EXPO ULL doesn't change the fundamental performance characteristics of X3D versus non-X3D CPUs, so while you can certainly pair an X3D chip with an EXPO ULL kit, G.Skill tells me that you're not going to see as large of a difference in performance from that pairing as you would with a non-X3D chip. That's why AMD is touting the performance gains of EXPO ULL with a Ryzen 7 9700X and not the Ryzen 7 9850X3D you might expect.

G.Skill also told me that EXPO ULL-ready memory requires stricter binning of individual memory chips during production, so it isn't just a software change that can be applied to existing modules. The company says the extra work involved in this stricter binning process means that modules supporting the feature are likely to be more expensive than kits that haven't undergone the same characterization.

Overall, then, EXPO ULL is likely to be a premium (and somewhat niche) addition to the EXPO program instead of a broad replacement for non-ULL EXPO profiles. Demanding gamers who need the lowest memory latency for the best performance in CPU-bound gaming scenarios will likely want an EXPO ULL kit regardless of the type of Ryzen CPU they're using. But we'll have to see just how much extra cash these kits demand in today's already eye-watering memory market and what benefit they have, if any, for AMD's massively popular X3D chips.

As the crowds in Taipei thin out and the crowds begin to disperse, our team on the ground at Computex 2026 ruminates on their experiences of the show itself and what it might mean for the industry at large. Day four will be our final entry into the Tom's Hardware Unfiltered series for Computex 2026, so we hope you've enjoyed peeking behind the curtain to gain an insider look at exactly what we've been up to this week.

Starting at the end is just wrong, so if you haven't yet caught up on all of the coverage coming out of Computex, be sure to check out the entire series of blogs.

• Computex 2026 Day 0
• Computex 2026 Day 1
• Computex 2026 Day 2
• Computex 2026 Day 3

Paul Alcorn: Editor-in-Chief

Day four was yet another hectic affair, with a string of meetings in the early morning continuing into the afternoon. This Computex has certainly had more attendees than I ever recall on the second day, and today was no different; there were a surprising number of people cramming all the aisles, and of course, jamming up the booths, which isn’t great if you’re trying to take pictures.

I also noticed that this Computex has far more business-to-business (B2B) focused companies in attendance. In the past, we typically saw strictly consumer products and the myriad of companies that feed into that ecosystem. This year was definitely an explosion of AI and data center technology, which is surprising. Also, the big showcases dedicated to data center hardware were absolutely packed, often just as busy as the ROG and ASRock booths of the world, if not busier. Others have noted that there really isn’t an Asia-based trade show for data center tech, and it seems that Computex has now become that destination at an incredibly fast rate.

Joe Shields: Staff Writer, Components

Day four is in the books. I managed to get about two to three hours of sleep last night, and it’s been a long day, even though I got all the meetings done early. Today’s journey began at the Grand Hyatt Hotel (Hyte), right next to the beautiful Taipei 101. Soaring over 1,600 feet in the air, the blend of traditional Asian aesthetics and modern engineering is a sight to behold.

Hyte showed off a few new items, including the Y50 case, a less expensive version of the popular Y50. We finally made it to the Gigabyte booth at the convention center and had a chance to see the X870E Infinity Next motherboard in person, and wow. It’s absolutely stunning with the 3D-printed metal heatinks and that lava-rock-like pattern. Be quiet! Showed off a few items, but I was impressed with the Light Base 803 chassis, and a new power supply (Dark Power Pro 14 IO) with software monitoring that even shows how much it costs to run your rig.

My last appointment was at Thermal Grizzly, who showed off new coatings for their waterblocks, new thermal pastes, and different versions of the WireView Pro for you RTX 5090 owners. By that point, I couldn’t tell if I was coming or going, and thankful all of my appointments were done. A little nap at the hotel and I am back to writing about Computex for the last night.

I’m incredibly thankful for the opportunity to come out to Computex. For me, it’s a completely different experience from CES in Las Vegas. Both have their positives and negatives. Tomorrow I’m excited to get a tour of Asus HQ, but really looking forward to starting the long, long trek back to the U.S. It was real, it was fun, but it wasn’t really fun. Ohio, here I come!

Jake Roach: Senior Analyst, CPUs

This was my last day in Taipei, and I tried making the most of it. The majority of my morning was spent running back and forth between the two halls of Nangang and talking with various companies, racking up some 20,000 steps in the process (the halls are literally across the street from each other). Some of the details of that tirade through the trade show I can’t talk about quite yet, but as the blisters slowly forming on my feet will tell you, I kept myself plenty busy.

My afternoon was spent at Asus HQ in the Beitou District (about an hour-long trek from Nangang on the MRT). I usually visit the Asus campus, and it is beautiful. There, I met up with several Asus reps and various other media to learn more about Asus’ announcements, as well as some products that are coming down the pike. But, as you might expect, I can’t talk about those quite yet.

Leading into Computex, the week always seems so long and grueling. And it is, make no mistake. But the irony is that right as I start to get adjusted to the 13-hour time difference and begin feeling like myself, I’m packing my bag to head home. It’s always too much time, but never enough. Oh well. I’m finishing my night with a nice plate of omurice, which I can not get at home, and trying to catch a few hours of sleep before spending 18 hours in a plane seat tomorrow.

Jeffrey Kampman: Senior Analyst, Graphics

After four intense days writing into the early hours, visiting vendor showcases, and criss-crossing the show floors at both TaiNEX halls, today was my last in Taiwan, so I once again headed over to Nangang before my departure to ferret out any last hidden hardware gems that we might have missed.

I walked through the massive Gigabyte booth with Joe, admired the company’s classy wood-trimmed Aero cases and Infinity graphics cards, and mused with Gigabyte staff about how the show and the industry have changed since I was last at Computex almost ten years ago.

DIY PC building is, for better or for worse, increasingly a guided, safe experience. The big component companies are now more than happy to serve you a full menu of coordinated parts that are practically guaranteed to come together into a coherent build. It’s getting harder and harder to find examples of earnest whimsy among booth after sprawling booth of Vera Rubin NVL72 racks and their supporting infrastructure. (Indeed, Vera Rubin NVL72 is so tightly standardized that racks from different server vendors are practically indistinguishable from one another!)

But my next stop was down the escalators to G.Skill’s booth, which maintains much of the freewheeling enthusiast spirit that I recall from my last visit to Computex so many years ago. If you want to see extreme overclockers chasing world records in real time through swirling clouds of liquid nitrogen, G.Skill’s booth is the place to be. And I got to go deep into the weeds with the company’s reps about DDR5 sub-timings, the resultant memory latencies, and their effect on modern CPU performance. After a week of parsing bold visions for the future of computing, it felt good to get back to the basics. Farewell, Computex, and I hope to be back in Taiwan soon.

When not being spotted at night markets or meeting crowds of adoring fans, the hardware industry’s biggest celebrity, Nvidia CEO Jensen Huang, spent most of his time at Computex this week making the case that computing as we know it is collapsing into one repeatable pattern built for AI agents; a blueprint that now runs across the cloud, the PC, the car, and the robot.

"There's a new computing pattern," the Nvidia CEO told reporters at a press gaggle the day after his GTC Taipei keynote, describing an agent architecture he calls a harness that orchestrates reasoning, memory, and tool use the same way whether it sits in a data center or a laptop.

He tied that claim to every product Nvidia detailed at the show, from the Vera data-center CPU now in full production to RTX Spark, its first Windows PC platform, shipping in laptops this fall.

One pattern, every machine

Huang told the room that he repeats the same keynote structure on purpose. "Every time I give you a keynote, it's like Top Gun 17, and it's exactly the same architecture," he said, "because I want you to know that the future of computing is this." The pattern begins with training and inference in the cloud and pushes outward to everything else: "Every edge device will become autonomous. Every edge device will have agentic systems."

He ran that blueprint through self-driving cars, humanoid robots, Nokia base stations, and imaging satellites, casting each as the same agent profile on different hardware. Curiously, the self-driving car got quite a bit of airtime, with Huang describing Nvidia's Alpamayo driving stack as a system that reasons in language rather than reacting to images, one that could read a "skill file" and watch a tutorial video to operate unfamiliar machinery the way a person would. "That's how autonomous vehicles are going to work in the future," he said. "It's essentially that agentic computing pattern with a physical AI model."

A CPU that generates tokens, not cores

Vera, on the data center side, is an 88-core Arm processor that Nvidia is now in full production with, pitching it as a chip built for agents rather than human users. "We built Vera for agents to use," Huang said. "Until six months ago, there were no agents, so that's the definition of a $0 billion market."

A hyperscale CPU piles on cores because humans lease them by the hundred, where an agent, Huang argued, "doesn't want to rent the CPU core, the agent wants to generate tokens." That pushed Nvidia toward single-thread speed and memory bandwidth over core count, and Huang claimed Vera offers the largest step up in single-threaded performance he has seen "in 25 years." His reasoning ties back to latency: "Humans are more patient than agents. Agents, they're working at nanosecond scale, not second scale."

Nvidia claims 1.8 times faster task completion than x86 and a 1.5 times instructions-per-clock gain over its Grace predecessor, with a 256-chip liquid-cooled Vera rack it says reaches six times the throughput of a conventional CPU rack. The chip ships on the back of nearly 2.5 million Grace units sold, and Anthropic, OpenAI, xAI, ByteDance, CoreWeave, and Oracle are named as early customers. CFO Colette Kress told investors on Nvidia's latest earnings call that the company sees "nearly $20 billion in total CPU revenue this year".

Phoronix's first public Vera benchmarks in May measured it roughly 10% ahead of AMD's 64-core EPYC 9575F and about 55% ahead of Intel's 128-core Xeon 6980P across selected Linux workloads. Nvidia ran those tests on pre-production silicon at its own headquarters, limited them to workloads it considers relevant, and, by Phoronix's account, switched off CPU power and frequency monitoring for the session.

Reinventing the PC after 40 years

As for RTX Spark, Huang says that it’s the first real rethink of the PC in four decades. "We have an opportunity after 40 years to go reinvent it for the age of AI," he said, predicting the machine shifts "from your PC being a tool to now really your PC being your system." He pushed even further: "Your laptop is going to be your R2-D2."

The top RTX Spark part, internally N1X, pairs a 20-core Arm CPU built by MediaTek (10 Cortex-X925 performance cores and 10 Cortex-A725 efficiency cores) with a Blackwell GPU carrying 6,144 CUDA cores, up to 128GB of LPDDR5X unified memory, and a 600 GB/s NVLink-C2C link, all on TSMC's 3nm node. Huang justified these specs with the same impatience he applied to Vera, arguing that an agent driving the machine won’t wait, so the software it touches, from Adobe to Blender, "cannot be slow."

The platform is launching in a market that Qualcomm had effectively dominated until its Windows on Arm exclusivity with Microsoft lapsed. Fall 2026 laptops are confirmed from Microsoft, Dell, HP, ASUS, Lenovo, and MSI, with Acer and Gigabyte to follow, and Nvidia says anti-cheat engines, including Easy Anti-Cheat and Denuvo, run natively on the chip. Asked why Nvidia would enter a low-margin business it has steered clear of for years, Huang said, "We don't really have to choose. The real question is, can we make a contribution?"

Vera's 88 cores are Nvidia's own custom Olympus design, its first ground-up server core since the Denver and Carmel projects, while RTX Spark's 20 cores are Arm's off-the-shelf Cortex reference designs licensed through MediaTek, one of them already a generation old. Huang's "same pattern everywhere" runs, at the silicon level, on two different CPUs.

When asked whether the Olympus cores would come to Windows PCs, Huang declined to commit. "Our preference is to use off-the-shelf cores whenever we can, because Arm also builds good cores," he said, adding that Olympus was pushed toward single-thread speed in a way standard many-core Arm parts weren’t: "We wanted to push single-threaded performance as far as we could push it." The first PC chip using Nvidia's own cores isn’t expected until 2028. Meanwhile, Morgan Stanley estimates Vera at around $5,000 per socket inside a vertically integrated rack.

What about memory?

DRAM contract prices have climbed sharply through 2026 as makers divert wafers to high-bandwidth memory, and Nvidia remains short of supply even as it locks in capacity, by Huang's own account: "We have enough supply for very robust growth. However, we are supply constrained."

"One of the best ways to improve memory use is to use extremely, extremely low precision," Huang said, pointing to NVFP4, Nvidia's 4-bit floating-point format that scales between four, eight, 16, and 32 bits and roughly doubles the parameters that fit in a given memory pool, the trick that lets RTX Spark hold larger models in its 128GB. He paired it with neural texture compression that cuts game texture memory by up to eight times in Nvidia's demos. At SK hynix's booth during the show, Huang signed an HBM4E wafer with the words "Please Make More."

Tucked in the corner of the ID-Cooling booth here at Computex 2026, a brand called Wokyis is showing off some seriously striking Mac Mini and Mac Studio docks. Its M5 Macintosh dock for the M4 Mac Mini is already for sale on Amazon in 10 Gbps and 80 Gbps flavors, offering retro aesthetics and a 1280 x 720 display, along with a spot on the back for an M.2 SSD.

• 10Gbps M5 Retro Dock Station for Mac mini M4
• 80Gbps M5 Retro Dock Station for Mac mini M4 Pro

But Wokyis also showed off NES-themed docks for the Mac Studio and Mac Mini, which are expected to head to Kickstarter in mid-July. The G7 Retro Mac Studio dock has a larger 7-inch flip-up screen (sadly, still 720p) behind NES-themed controller buttons on top, with the D pad controlling volume, screen brightness, and media playback, and the B and A buttons handling copy and paste duties. The “Start” button appears to turn off the screen, while the “Select” button takes a screenshot. Like the M5, the G7 will come in 10 Gbps and 80 Gbps (Thunderbolt 5) variants, support an M.2 SSD (up to 2280) in a slot under the screen, and up to 36W of PD power. The Thunderbolt variant will support up to 8K video at 60 Hz, via DisplayPort and HDMI.

Aside from the card reader slots on the front (including 1 Gbps CF Express), you get four 10 Gbps USB ports (2 each, A and C), and three or two USB 2.0 ports (one fewer on the 80 Gbps model, according to the company’s specs table), along with an audio jack. The Kickstarter notification signup page promises a $199 MSRP, with an early discount of 45% ($109). Presumably, that’s the starting price for the 10 Gbps. I’d expect the 80 Gbps model to have an MSRP of around $349 or more, as the smaller-screen 80 Gbps M5 sells for $339 on Amazon.

As a nerd of a certain age, I’m all for this kind of old-school hardware nostalgia paired with modern computing hardware. I have to wonder whether Nintendo’s lawyers might take an interest, but Acermagic did a similar thing with its X5 mini PC earlier this year, and it was on sale on Amazon for months (though it currently seems to be sold out everywhere).

So if you’re interested in one of Wokyis’ NES-themed docks, you might want to sign up to be notified when they launch. Just remember, as always, that pledging at Kickstarter does not guarantee you’ll get a product in the promised shipping window, or at all.

For a brief time from 2024 through 2025, it really seemed that 16GB of RAM was the new standard on mid-range and premium laptops. Microsoft made 16GB of memory a standard to be labeled a Copilot+ PC, and Apple made 16GB the minimum on all of its systems.

Then came the MacBook Neo. It was made of aluminum, looked and felt premium, and used a version of the A18 Pro system on a chip from the iPhone 16 Pro, which features 8GB of RAM.

While there was some blowback from the enthusiast community, most reviewers found that for the most basic workloads, 8GB is workable on macOS (though you'll likely run into swap and memory pressure if you overload it). For everyone else, 16GB is the standard on the MacBook Air and Pro.

But at Computex, we saw that the Windows PC industry is also moving to 8GB. The $699 Dell XPS 13, one of the standout products of the show, will start at 8GB of memory along with an Intel Core Series 3 ("Wildcat Lake") processor. Acer, too, has a $699 Wildcat Lake-based system, the Swift Air 14, which will have an Intel Core 5 chip, 512GB of storage, and, yes, 8GB of RAM.

Ahead of the show, Chuwi announced its UniBook, with a weaker Core Series 3 304 CPU, 256GB of storage, and, you guessed it, 8GB of RAM. That's set to be the cheapest one so far, with Chuwi promising a launch "around $449."

And in May, when Microsoft announced its latest Surface for Business devices, it included a new Surface Laptop for Business 13-inch with 8GB of RAM starting at a whopping $1,299.99, so 8GB isn't just relegated to the budget space.

That's not to say everyone is doing it, at least yet. Some Wildcat Lake laptops are sticking to 16GB. For instance, the $899.99 Lenovo IdeaPad Slim 3i has 16GB of RAM, though it's clearly more expensive than what Apple, Dell, Acer, and Chuwi are offering. (As of this writing, Lenovo's website suggested it's 25% off an "estimated value" of $1,199.99.)

One interesting thing to note: the Windows laptops with 8GB of RAM still have Copilot keys. This doesn't make them Copilot+ laptops, which requires Windows 11, the Copilot key, and 16GB of RAM, but hey, cloud AI is still AI, I guess.

Intel's Wildcat Lake website has a long list of partners with new laptops, and I'm sure more of them will have 8GB. And we know more is coming. The Acer Aspire Go 15, the first laptop with Qualcomm's upcoming budget-focused Snapdragon C, has "up to" 8GB of memory, according to Acer's press release, suggesting the possibility of even less memory. The minimum specs to run Windows 11 include 4GB of memory, after all.

Unlike some of my colleagues, I still believe 8GB of memory is workable for the right people (To be clear, it's not the route I would go). That's people who mostly use a web browser and light apps, don't use many tabs, and mostly do one or two things at a time. They do exist, and they should have options other than paying $1,500 for 16GB of RAM. We've seen that macOS can largely handle a premium-feeling experience with 8GB (and will hopefully move to 12GB with its next chip upgrade), so now it will be the Windows world's turn to prove the same.

But I also hoped the days where 8GB of memory could potentially bottleneck you were over. It is both unfortunate and ironic that it was the rush towards preparing devices for AI that pushed us away from a longstanding 8GB minimum, but it is because the price of memory to keep AI research going has pushed component prices so high that laptop makers have no choice but to go back so that people can afford entry and mid-tier systems.

If you thought we were past the days of 8GB systems, you're wrong. You don't have to get one (depending on what you do, perhaps you shouldn't), but for something affordable, at least there's an option at all. Maybe in a few years, the RAM crisis will have subsided, and we'll laugh about this on 16GB entry-level machines. But not anytime soon.

Qualcomm's Snapdragon C is making a major play for the hottest laptop market in 2026: The ultra budget segment. But competition is stiff, with Apple's MacBook Neo and Intel Wildcat Lake Windows laptops offering strong performance and battery life in affordable packages - even with sky-high global memory prices.

We sat down with Qualcomm's SVP of compute and gaming, Kedar Kondap, and other Qualcomm representatives at Computex 2026 to hear how Snapdragon C fits into its existing product lineup, and how Qualcomm considers itself uniquely positioned to offer a comprehensive ecosystem of agentic AI devices and software.

This transcript has been lightly edited for clarity.

Kedar Kondap, Qualcomm SVP of Compute and Gaming: Our journey was not about solving what has happened in the past 30 years of PC innovation, but solving what's coming next in PCs. A lot of the innovation that has happened, we've obviously proved to the market that we're exceeding performance. We focused on three big metrics all along the way. One was leading with performance and making sure that we have leading performance in all of our silicon. We want to make sure we lead with power. Obviously, that used to be something super critical across the board, even as we start looking at newer generation of devices, and third, as you look at AI as a key metric.

When you think about the keynote today, and what we talked, what I shall talk about, as we enter this agentic world, it is more and more important. Each of these things play a very important role, whether it's performance, whether it's power, or whether it's AI, and the ability to run these intelligently. Right from whether it is a very small device, all the way to the data center. You saw how we're innovating, and the PC is no different. So we launched our X series processors, we extended that to the X Elite, the X Plus, and the X family. We launched the X2 Plus, we launched the X2 Elite, and the X2 Elite Extreme, and we wanted to make sure our intent was very simple: we wanted to make sure that the experiences we offer are available to all the consumers at every price point that we could address. What we introduced yesterday is a new class of platforms, the Snapdragon C. Our intent with that is to address platforms in lower price points that we can go and make sure that we can deliver the same performance pillar, the same battery life pillar, as well as provide AI functionality to all the consumers at price points that were never heard of.

So with that, we wanted to make sure that we have a full stack of products, we are addressing the needs of what consumers want. Consumers want the best performance, the best battery life, and as we enter this world of agentic beta, we want to make sure that across different devices, we're addressing all of these price points. So, we’re excited to be here, and I know there's a little bit of a longer introduction, but I want to make sure you have the perspective of where we come from.

Journalist 1: Thanks, Kadar. Nicole. You were just on stage with Advantech. Can you get us a little background on your announcements and what was said on stage?

Qualcomm Representative: Advantech has been a great partner of ours. I've done more recently the industrial business for Qualcomm, and so, you know, for us to get into this new area, and industrial is actually changing very rapidly. We are starting to see AI enter into the operation, advantage has been here in this space for a number of years, and so we announced a variety of different products to them over the last couple of years. Great to actually be at their keynote. We also announced yesterday [unintelligible] robotics reference design, which is something that is a new area for us. We've been partnered with Advantech for a number of years in this space, but the robotic session design will be a humanoid focus session.

Journalist 2: I'm wondering, Nvidia, a few hours ago, announced they were entering PC market. You guys have had it yourselves for the last two years. How are you thinking about how others come into [unintelligible].

Kedar: Welcome to the family [laughs]. We are, you know, we're excited when you think about the investments that we've made over the last several years, it's a good endorsement to the fact that there is an ecosystem that's growing outside of x86. We invested early on, we invested many years ago, with right from whether it is driving the ecosystem, driving the entire platform story. Whether it is getting the printers to work, whether it's getting the software apps to be compatible, whether it is getting the docs and peripherals to work, whether it's getting more than 2500 games to be compatible with Snapdragon, we led the way in driving that ecosystem, and I think this is positive tailwinds for the entire ecosystem. They'll tell us how we're all taking the ball forward in the trajectory that we started.

Journalist 3: This is for Nichole. She's had remarkable success in automotive and moving into robotics. Is there anything you’ve learned [unintelligible].

Qualcomm Representative: We’re learned a lot from automotive. I think AI at the edge with robotics is a really fascinating space, because you are starting to see this transition where you start to apply AI to start off with an unskilled worker, gradually start to go build up on level of skills, and ultimately start to get to something great. A lot of the underlying capability is similar to what we saw in the early days of automotive, especially the mobility part is quite similar. That will scale very quickly. We will start to see coarse dexterity, so that it's essentially transportation of goods, et cetera. That will happen over the next couple of years, and then the more complex final precision tasks will take more time.

But we are betting across the board, we are betting on models, we are betting on full embodiments. You'll hear more about this tomorrow. We're betting on a variety of different form factors with arms as well. We are also starting to look at what we can do in the end effector space, so we have a lot of technology around precision for the actual and effect of the arms, the actual digits, and what it is. What I like about robotics, which is quite different from the car, is that within the same embodiment, you have to have a lot of different technologies that cooperate, and that is something that we have a lot of capabilities. Lot more to come, but I think super interesting.

Journalist 4: When I was at the keynote from Jensen Huang, I was surprised when he mentioned the RTX Spark platform will support every Windows software ever written, and you suffered from Windows software in running on ARM. So, could you comment on that? And another question, if you could maybe share some more details about Snapdragon C platform, especially the TOPS. I saw a model from Asus, but it was behind in the glass box, and they didn’t have a lot to say of interest in that matter.

[Jensen Huang] mentioned that they will do an announcement tomorrow with Microsoft about the RTX Spark, and he mentioned that every Windows software that has ever been written will run on their platform, and this is not true for your platform, and probably for any Windows on ARM platform right now. So, I'm wondering if you want to comment on that.

Kedar Kondap: I don't want to speculate, but I'll tell you, our partnership with Microsoft has obviously gone several years in the past. We worked with them very closely. We launched the first Copilot class pieces together. We launched the first platforms where Microsoft OS supported it, and supported how these distributed computing work across a different course. So maybe once we start getting more information, we can, but I think I'm sure the engagement of Microsoft is strong enough where we work with them to build this entire ecosystem to make sure that it's compatible with Snapdragon and the architecture, so maybe once we get more information and get more, we can look at it.

The second question on Snapdragon C, so we haven't yet given out the specs for the products, but I'll tell you how I think about Snapdragon C. We wanted to, our OEMs are very anxious to bring this product to market, as is Qualcomm. There is, as we all know, there is a memory supply challenge in the market, and I've heard there's a storage supply that's also challenged, and we wanted to make sure that we have an offering where we can address a lot of price points that we've never addressed before, so think price points below what Snapdragon X has offered in the past. Snapdragon X will give you its relatable, you know, it's great platforms that get 599 today, some of them that hit 500 and above the X plus goes above and X Elite goes above that, so think it's a tier below what we can offer with Snapdragon X, and so our intent with that is to drive the same level of capabilities, obviously scaled to that tier, and what I'll tell you, it's just like the way we've exceeded expectations in launching specs of our products in the X series family of products. You should expect that the C in its class of products will lead its way, but we'll obviously give out specs for you.

Journalist 5: I've got a question regarding [Qualcomm’s] vision on how the tokens in a car, in a PC, in a smartphone and adapts and talk to each other. Cristiano said that we're going to need zillions of tokens, and they're going to be orchestrated. It means different parts going to generate the token, where it's relevant. Okay, like latency, horsepower. How do you gonna orchestrate them with your industry? Are you planning to build a software ledger, SDKs, or building a patented or open system? How do you plan in this vision to open everyone with a heterogeneous chip market, it's going to be difficult to make everyone work.

Qualcomm Representative: So I think it will vary by the ecosystem, but yeah, we will certainly build orchestrate the [unintelligible] where it does make sense. I'll give you an example of a car. We had already seen in the car a tremendous amount of content. You don't actually need, you can't even expect in many cases to be running models that all just have to go back to the cloud, just because of networking latency. So we have deployed already earlier this year 30 million parameter models in car. It depends upon what the types of use cases you're trying to run, but then if you think about this in the context of splitting search versus what it is that you're trying to run locally within the vehicle that is happening today. I think this starts to get more and more sophisticated as you start to define what is the use of the model at the edge.

So, with industrial, for example, we are starting to see already VMA is getting deployed in cameras when you have evidence at the edge, where you can annotate at the edge, send it out to cloud as an input, so it's going to depend quite a bit on which the ecosystem that you're trying to run your account. If you look at use cases like polling, if you look at use cases like search, those are probably obviously much more in consumer and enterprise nets, which should have different network orchestration, probably much more controlled by traditional ecosystems. An orchestrator is going to be a fairly standard offering as well.

Kedar Kondap: Aiden. I'm going to add a little bit more context also to what we'll just say, you know, the way you should think about how we think about overall income advantage, it's going to be the context awareness that you get across the various devices, and think of a personal knowledge graph that you build in the examples that he showed, right from whether you're wearing an XR glasses. We are able to connect. You can see what you see, you hear what you hear. You can have context awareness of what the user is doing. It's going to listen to what information you talk to when you're, say, at the doctor's office. You understand exactly when your next appointment is.

The ability for it to abstract all of this information, transfer that knowledge graph agentically into your calendar, into what your daily routine is, be able to take your personal information, whether it's talking about your healthcare routine. The goal that we see is first is to bring all of this orchestration together, and Qualcomm is uniquely positioned, as you can tell, right from whether it's a small physical personal device, whether it's a ring, or whether it's an XR glass, all the way from phones to PCs to automotive to robotics and data centers, we feel like we're in a unique position like one another.

Second, on the question about running stuff on device, now we've always said the world in the last two years. You know, I remember when we launched Snapdragon X Elite. We said with a lot of pride that, oh, we can run 30 billion parameter models. You know, today, fast forward that conversation, we showed 20 30 billion parameter models running on the C Snapdragon X platform. So, models have evolved. I can, what I could do earlier with, you know, accuracy of quantization and stuff, I can do a lot more than I couldn't do in the past. So the industry is evolving, where we're innovating, we're adding more capabilities to each of these devices on-prem, I'll say, or physically on device. At the same time, we know that the token economics, as Krishna showed, everybody in the industry see it?

So, if any of you use, if you have a poster subscription towards AI, you know you run out of tokens very quickly, and there is the fatigue is real, the fatigue on the side of a consumer, the fatigue on the side of the model. You cannot have both of those, the balance doesn't exist today, so the way we see it is one, you can connect all these orchestrations across the devices, same, you can orchestrate on what runs locally on the device, and of course we believe some of it will go in the cloud, and that hybrid orchestration is where we believe the industry is going, so build the knowledge graph across devices, run what you can locally on the device, if not, if the model is large enough, it will go to the cloud, and this whole equation will evolve over time as models start to become smaller, as they start getting quantized accuracies to be better.

What was, like I said, a 30 billion parameter running on a Snapdragon X, I can already run 30 billion parameters models quantized with very good accuracy. So that's how what we mean by this industry and ecosystem is going to change.

Journalist 6: I would like to touch on Snapdragon C again. You expect the platform to be a regional-specific solution, like for emerging markets, for example, and Qualcomm has great experiences in markets like India, for example, where do you expect it to be a global platform? And I'd like to touch on the NPU as well. Previously, every Snapdragon X platform, at least, had an NPU that OEM Certified Assistance for Copilot Plus seems to be the first solution where you sort of loosen your own set of requirements for a Snapdragon compute platform, and what led you to that decision?

Kedar Kondap: So, Nicholas, first I'll answer your first question, which is: it is a global platform. You will see this device launch globally. You know, for us it's there's a large stamp, as you know, in this particular segment. Lot of consumers that use PCs that sit in, I'll say, below the $500 price point. So, we have a large stamp that we can address there. The TAM [Total Addressable Market], as you know, is naturally biased towards emerging markets. So, from that perspective, yes, the focus, you'll see the platforms launch in many of the emerging markets, as well as developed markets. The TAM is much reduced in that, so it's just a function of the definition of where the time sits.

Your second question around NPU, no the platform does have an NPU, we just haven't talked about the sizing, but as you can tell, the silicon economics, we're sizing everything with the capabilities to be able to run use cases synchronously with the price point that we're addressing. So, think of it as you'll still be able to run, get a lot of the capabilities, you'll still be able to run a lot of use cases we talked about, while preserving the performance and battery life goodness that we bring with the Snapdragon product, so you'll see something very similar.

Journalist 7: There's been a lot of good questions on Snapdragon C, but if you're not competing with Neo, what specific product lines of products are you competing kind of in that below $500 price range?

Kedar Kondap: You know, you already heard a couple of our OEM partners talk about devices. I think Asus talked about the device. I want to make sure I don't jump the gun with the excitement that our partners want to launch it, but I think Asus has already announced. We've also addressed the fact that Lenovo, as well as HP, we announced with HP. Just want to give out somebody's information. We've announced with the OEMs, it will sit as I said, you know, Snapdragon X sits at around 599 type products, and it competes, as you know, better than what you can get with Neo and performance insights.

This gets the price points that it would be below 500, obviously. Now, the thing I always have to caveat is the memory and storage prices. I never know what it's going to do to devise price points, but that's how the platform is positioned. Think of it as small core competing products, but you know I'm nervous to say that, because it's going to be so much better than what existed in the market, so that's why I don't want to tell you that it's compete against that, because it's a different class, it's going to be a lot better than what's what you're used to seeing in market,

Journalist 8: Sorry to ambush you with another Nvidia question, as I know you already said you didn't see the announcement. They are going after a market that you guys have historically not went after, I think a few years ago you publicly said this is just not our wheelhouse, so are you guys planning to compete with what Nvidia is doing now, and if so, how?

Kedar: Like I said, I haven't seen this. I don't know what they've announced, but at the end of the day, look, we've come a long way from where we started. We go a couple years back, what we said, Rich, was we have the legacy, we have the technology in even things such as gaming. We come from, as you know, we have a very strong game studio house. We work with all the game engine guys to work with the net engine guys. We work with the studios on the mobile side and other platforms that we work with, but we bring all of the games to Snapdragon. Since we launched Snapdragon X in less than 24 months, we've done a lot of games that work with Snapdragon effectively go from 1300 to more than 2500, 2600. So I think what we've mentioned in the past is not that we couldn't support it, what we've said is we want to carefully not position this as a gaming laptop, right? The gamers who think of launching this as a AAA game-based laptop is not what we want to position.

We don't want to create confusion with what we're addressing, but from a technical perspective, nothing prevents us from addressing a lot of that, because, as you know, even when you think about creators, we've talked about our partnership, and what we've done with Adobe, what we've done with Black Magic, and others. So, we've already showcased that the entire creative industry is something that we support. We sold multiple use cases, so it's the entire ecosystem that already we've been addressing. So, like I said, I'm assuming that their introducing their platforms in market today is tailwinds for what the ecosystem will see as a strong showcase of non-x86 architecture.

Journalist 9: On the question of the Snapdragon C, somebody spoke about the emerging markets. One of the things which has happened in those emerging markets that arise on the tablet usage. So in the past three years, four years post pandemic, you can say that many of the people are buying who are stuck between a rock and a hard place, that you know, laptops are slightly expensive, they have a full use case of the productivity of a laptop, but they wanted a slightly bigger screen, so they are buying with sliders and the keyboard, especially the emerging markets. So, I just want to ask you, is your target that consumer, and do many of those cohorts is parents buying for the kids or education market, private or public, both? And the public market, you know, you know some of the PC loans, they've really done well, and they are restricted that success in the past year or so in education and some of those emerging markets, because the price points are creeping up now. So I just wanted to double click on that. Who is that target consumer in those emerging markets where we expect the volume to come from?

Kedar Kondap: So, as you know, we already play in a very strong manner with the Android tablets in all of those ecosystems, we have very strong partnerships across almost all the OEMs that you can think about. We have Samsung, OV, WOSU, Lenovo. We have all of these tablets that have launched. There are specific tablets I've launched that are focused on gaming all the way to productivity, all the way to education. So we have very strong portfolio products launched even in emerging markets with our partners.

With Snapdragon C, I can see a market out there, as you correctly pointed out, education being a very strong segment would address all of that. So, if you saw the press release that we talked about, which Snapdragon C we specifically call out that we will be in the education space. We are doing a bunch of pilot programs with our partners to go and address that. I wouldn't necessarily say that it's going to replace Android tablets, necessarily. I think hard to call whether that ecosystem is going to move away from Android to Windows, but right now they coexist pretty well. Like, there's a good TAM that's available for Android tablets with a stylus, as you correctly pointed out, as well as a TAM with Windows PC, and I think with this particular one, for now we'll be launching the Windows segment.

Journalist 10: Can I talk about we just launched, like Qualcomm with Asus A16? Will you deepen with Taiwan cooperation?

Kedar Kondap: I can't tell you how grateful I am for the partnership with the entire Taiwan ecosystem. The partnership goes obviously deep partnership with the ASUS, and we're grateful for the partnership with ASUS, brings a lot of innovation and market. If you've seen, if you haven't already played around with it, or if you haven't seen it, I encourage you to look at the A16 device with the Snapdragon X2 Elite Extreme product. It's a beautiful, thin, light laptop with significant hours of battery life, so it's a, it's a beautiful laptop, and that's because Asus brings a lot of innovation. You heard yesterday from Acer with the Snapdragon C platform that they're launching, but it goes beyond just the OEM partnership for us here in Taiwan, right?

We work very closely with all the BIOS guys, whether you know all the IBDs, we work with the inside, we work with AI devotees, we work with all the partners there, we work with all of the EC manufacturers, we work with ecosystems, so it's not just one cookware, we work with the camera sensors. So we do a standard ecosystem partnership summit here in Taiwan, the offense is just because innovation is going to happen here, and I think what you hear, even as I speak on Wednesday, is as we talk about what we try to throw the vision of moving to agenting, it's going to need a lot of innovation, that innovation is going to change the way the PCs are going to look, is going to change the edge appliance market, is going to change, because now you're going to be running these hybrid models, running and stuff, and we really believe that Taiwan is the hub for driving innovation, and, like I said, we're very grateful for the partnership that we've had with this ecosystem for the last many years.

Journalist 11: Looking towards the intelligent orchestrator for managing workloads. Do you see a primary device managing the personal AI agents? Would it be a smartphone, a transportable puck, a locally hosted cloud instance, for example?

Qualcomm Representative: I think the orchestrator is actually going to be associated with the user, so it's going to be a personal orchestrator, and the devices will evolve along that orchestrator. I think it depends upon the specific ecosystem that you refer to. So, let me pick an ecosystem that may be different from a personal device that is owned by an Apple or Google system. There are several ecosystems where the personal draft, for example, is specific to the enterprise that would like to own that specific data. It could be within an industrial environment, it could be something that requires the data to be resident within the premise owned by that specific enterprise.

In those environments, it really just comes down to what is the type of data that is being exchanged, whether it is data at the edge, data that is network specific, that is manipulated, that is interacting by the specific data. As you start to think about this in the context of the enterprise, I think, depending upon the type of data that this might be, so if it's a personal device like a camera or my glasses, then it starts to move towards the context within the data, there will be changes in the interaction, so for example, today if you think about enterprise data that is sitting on your laptop, that is part of your email, that is part of your SharePoint. How does that interact with the personal devices that you have? Those are things that, in my mind, are still going to get sorted through, but you're going to start to see the personal graph nature move more and more towards the user itself as a personal device.

We believe that the cross-device, as I said earlier, is what will play a super important role. The context awareness from the agent having access to your personal information and bringing that context to information from one device to another, and being able to drive that continuum, is what's going to differentiate more common. We feel we're in a very unique position to be able to go from pocket all the way to the cloud in driving that entire ecosystem.

Journalist 12: I have a question about the economics of the new world you mentioned already, the memory crisis, the skyrocketing memory crisis. I would like to know what impact you expect of the rising memory prices on the development of these agents in future on the consumer side, enterprise side, but also maybe on the other platform side.

Kedar Kondap: I think let's separate out two different things. One is we've all people that have been in this industry long enough to know the sequential nature of what happens with memory. Of course, this time is an anomaly in terms of what we're all seeing. I respect that, but at the same time, we know that eventually the supply chain economics balance out with that, right?

Like, over time, it may not come to the same levels as what we've seen in the past, because this is obviously a super cycle in terms of memory that we've never seen before, largely because of the needs of what AI, what the ecosystem needs. At the same time, though the token economic problem is a real problem, which is you as a consumer you want to be able to use more AI and you want to be able to be more efficient with what you're trying to do, but you're there's only so much in terms of affordability, same thing on the enterprise side, all of what you think about op-ex moving to capex, you have to be able to have that migration happen where you see your op-ex that you invested in and you're going to start moving to capex, or you're going to start getting every employee is going to get multiple models and drive that efficiency that you can get as an enterprise that you want, that migration is going to happen once you move to the other side.

With that migration, the entire equation of economics changes. Why? Because now, if I'm an enterprise, I know that running stuff on device… I'll give you an example, if I can now start running 5070 100 billion parameter models locally on prem, whether the device is a laptop, whether it's a 5 billion parameter model that runs on a phone, or whether it's an edge of line that sits in my desk running a 50-100 billion parameter model, or it's an influencing card that sits in on-prem, I know that the token economics dramatically changed, but I also know that as an enterprise to be able to run trillion parameter models, I'm going to go to the cloud, but I know I need to invest to be able to run on device.

So this balance of equation is what's important, and we feel like this is where we've always said the world is going to go hybrid, we've always said it's not once you're done with training in the cloud, it's going to start moving to the edge with your inferencing, and we feel very good about the position we're in, because that's where we've been investing for the last several years to drive this on device as well as hybrid approach.

Journalist 13: While wireless transition is a very big effort [unintelligible] has put a lot of time in those transitions and a strong focus, you almost single-handed move the industry to the next level, and that Qualcomm changed a lot, now it's more, much more diversified. So, my question is, is Qualcomm from today able to have the same focus on the safety transition in the future, or it may be more help from the industry?

Qualcomm Representative: We are wireless company first, so yes, we have a lot of focus on safety. I think wireless is in the DNA of the company, and wireless is really part and parcel of everything that we do. I mean, if you think about the complexity of building a cell phone, then there is a reason why very few companies are successful with these things. I think the 6G transition, maybe just to speak with 6G, is a very interesting transition, because we believe that the networks are going to actually become very intelligent in 6G, you'll be able to get a sense of creating a virtual digital trend of environments around you. It will bring telcos into the fore in terms of being able to get a much better sense of the physical world. It's a major area of work for us, and these G transitions take a decade, right? So you know that it took a decade for us to be develop 5G.

Journalist 14: Kedar, you answered a couple questions already on the personal graph and that comment about the agent moving closer to the user. I get the point that Qualcomm spans across multiple devices, and that they're, you know, that you're at a unique advantage because of that. But can you take us through how that would be architected? I guess inevitably, if it's spanning across devices, cloud has to be involved in this process, right?

Kedar Kondap: I think Brian, the way you should think about it is, first, we know that the entire ecosystem is fragmented, and we believe that as different hyperscalers, as different model providers, as different OS vendors, as different silicon vendors, all of these need to come together. We were not giving out much information today on what we're doing there, but as you can tell, what we are trying to indicate to the industry is today we have the ability to thread all of these things together, and we're uniquely positioned on how we want to do this over the next several months. We will come out more with respect to how we want to be able to tie these together, but the industry challenge that you highlight, and that's why we feel like we're at the center of driving innovation across this to bring this industry along to try something that's innovative.

Journalist 15: I'm going to be one of those guys who brings it back to the ARM thing. I'm just going to read you what Jensen said in his keynote, because I think I'm just going to read it deadpan. “This computer literally runs everything the world has ever created, and it runs agents.” So I don't know if that's true, but I kind of wanted to give it to you, and kind of hear, what are you guys working on in terms of like compatibility, or even emulation, because we've seen a lot of interesting ARM-based emulation coming from like open source places, so is there anything you guys are working on in terms of compatibility that is trying to push things forward?

Qualcomm Representative: I don't know how to use a broad statement like this to give you a full answer, but I'll tell you what we'll do today. There's about the last night track with my team, was we were worried about 50 claws that are available in my team, and from a snap back in perspective, X series perspective, these are run on the device, I think Krishna showed a bunch of these, a bunch of claws that he showed on stage, so you should imagine that we're, we've been leading the industry with driving agentic AI and orchestration on the PC for the last couple years.

So a lot of these models already exist, a lot of these claws run very effectively on Snapdragon, the models are running effectively on Snapdragon, so you should just assume right now that we've shown you the data that we're already dealing with industry with tight innovation.

Journalist 16: Do you see robotics requiring the two layers of intelligence like with autonomous vehicles, or will the user interaction be more integrated with its physical functionality? Essentially, will the robot-human interaction be controlled more on the user side or the robot side?

Qualcomm Representative: I think that is going to predominantly be very similar to what they can do today already. So, language in the primary interface, what the human is talking about. Where it starts to get interesting is when you start to get into responding to a command. So, for example, if the human would like the robot to go do something that kicks off a task for the robot to be able to complete, and that is usually not a question answer conversational type interaction that usually requires the robot to be able to take on a longer horizon task that brings in additional models that bring in additional tasks that are outside of what humans might typically engage in, but the primary interface will remain the same conversation.

Journalist 17: I'll bring it back to Nvidia. I just want to add that from a CPU and NPU perspective, I think, from a GPU perspective the RTX 5070 class, what I have been missing is like what is called Qualcomm’s view from the GPU premium side of the market. Let's say personal computer market, so I need a bit of clarity there, and the second part would be to just focus it towards the developers and the AI community.

Kedar Kondap: Let me tell you, let me address your first question, which is when we look at launching a particular platform in market, whether it's the Snapdragon X Elite, the X Plus or the X. We always look at what the market needs, and we have a way to size where we feel like workloads are best run on a platform. It's the composition of how you could think about silicon. We have a very powerful GPU. We have the IP that we've invested in the high-performance GPUs and CPUs in-house for the last many, many years. We have to make sure that we're addressing a certain price point from X Elite to X2 Elite and X2 Elite Extreme. We size a significant improvement in our GPU performance, largely to address a certain set at the same time. You have to know that where we're shipping products are in certain price points with that target ASP in mind. I don't know what the industry is looking at with newer platforms are getting launched in market.

I haven't seen any of that data yet, but I'll tell you, all the cores that we support are sized to do that. With respect to their second question on developers, we've come a long way today in terms of where you see developers. Like, the entire developer ecosystem is behind what we've done. I talked massive numbers in terms of apps ecosystem, we talked in terms of developers porting apps natively on platforms, natively on Snapdragon, optimized to the NPU, so we have all the tools, everything that we've provided. So I think you should just expect that as we start moving into this new era of agentic…

Brian asked the question of how we're going to bring all these together, you should assume that we're working with the entire ecosystem, because I'll repeat what I said earlier. We are in a very unique position to be able to bring all of that orchestration, so whether it's a wearable, like an earbud or a watch, all the way to whether it's a PC all the way to whether it's a tablet, auto, XR, you name it. We're going to make sure that we're sizing all our platforms and technology to what the industry needs for that particular segment, so it's very segment-based in terms of how we look at the market.

[Session Ends]

One of my first stops on Day 4 of Computex 2026, I swung by the Grand Hyatt Hotel, right next to the 1,600-plus-ft Taipei 101 tower, to visit Hyte, who had a lot to share, much of it with value in mind.

The new mid-tower Y50 chassis, a slightly smaller version of the popular Y70, retains all the hallmarks of the Y-Series cases and modernizes the platform. The Y70 is now positioned as the more premium model with the vertical mount (and you can put a screen in it).

The Y50 offers users the same stylish aesthetic at a much lower price point, under $100. The Hyte Y50 RGB offers the same 3-panel glass and adds a step down for the side fans to allow more room for radiators. Hyte offers the Y50 in four colorways: Snow White, Pitch Black, Black Cherry, Strawberry Milk, and Taro Milk as options, the same as the Y70 Touch.

Internally, the Y50 supports up to EATX motherboards and rear-connect motherboards from Asus, MSI (Project Zero), and Gigabyte (Stealth), as well as large graphics cards. It comes with their new FA120 RGB single-frame fans on the side, and a single FA120 on the rear. Hyte is also launching the aesthetic cable accessory kit that matches their colorways for $69.99 in July.

You get a 24-pin ATX forboard power, dual 8-pin EPS connectors to power the processor, and one 12-V2x6 cable. Hyte also displayed a couple of new fans in the FA12 and the THICC FR12 RGB, their first ARGB fans. The new FA12 (25mm thick) comes as a single fan (FA12 solo) or as a unibody 360mm frame, and both traditional and reverse-style blades. The PWM-controlled devices run up to 1,500 RPM and output 70 CFM at 1.5 mm H2O static pressure alone, and 210 CFM at 033 dBA maximum, so they are pretty quiet. You can pick up the trio for $24.99 or a 4-pack of singles for $29.99.

The THICC FR12 Trio adds a unique 360-degree ring of frosted RGB lighting around all three fans, and as the name implies, is thicker at 32mm. The added depth increases static pressure to 4.4 mm-H2O, airflow to 57.14, and spin up to 1,900 RPM. It’s not a ton of airflow or pressure, but it should certainly get the job done.

You can pick up the Trio for $44.99, and the Solo for $19.99. If the RGB lighting from the fans isn’t enough, the ARGB Noodle LN80 noodle can match your vibe for $29.99.Also in the picture is a prototype AIO, which uses a unique, clear unibody (360mm) of fans. They aren’t sure if it will come to market (it depends on feedback), and it’s not finalized (square or circular block cover?), but if it does, it will be in 4Q this year, and they are targeting a $99 price point.

The last item from the Hyte booth is an update to the Nexus 2.0 software. Nexus 3.0 software was written from the ground up with new code and will be a web-based GUI with a much lighter footprint, using 75% fewer resources while keeping all the same functionality. The updated software also now works on Mac (see MacBook in the image on the left) and Linux, and connects to your phone (QR code). Their goal is to get a public beta out by Q3 and start transitioning users off shortly after.

Each spring, leading PC hardware vendors gather in Taiwan for Computex, a massive showcase of everything from svelte new laptops to brighter and faster monitors, cooling, storage, CPUs, and motherboards. If you love PCs and related tech as much as we do, this show is the most important one of the year.

This year’s show was dubbed “AI Together," so there was, of course, plenty of focus on that monstrous marketing term and all-consuming tech trend. But because of AI’s continuing hardware demands, which have driven up prices while driving down general availability of things like RAM and storage, we saw a strange mix of products. Some companies were aiming to make things more affordable, while others launched future-looking hardware aimed at new paradigms of personal computing and using new kinds of manufacturing, which definitely won’t come cheap.

We even saw AMD re-release a CPU from four years ago, which will likely push more PC builders back to old – but more affordable – DDR4. Combine that with the recent listings for new SATA SSDs, and your next new PC might be decidedly retro.

This year’s standout products include handheld gaming PCs that may break AMD’s dominance, the best Windows-based answer to Apple’s MacBook Neo we’ve seen so far, and a motherboard wrapped in 3D-printed metal with 64 power phases and a max current rating that’s not usually seen outside of industrial machinery.

Intel Arc G3 Extreme

Intel Arc G3 Extreme

Intel has tried to break into the handheld market previously, but the new Arc G3 range is its bid to finally make a mark. Intel describes the two SoCs as GPUs with integrated CPUs, hence the Arc branding, and that’s actually a good way to think about them. The company cut two P-cores, as well as slimmed down the Thunderbolt and display engine, specifically to target the handheld form factor. It’s really the first time we’re seeing purpose-built SoCs for handhelds, rather than rebranded versions of existing laptop designs.

But the idea of an iCPU really comes to life with what Intel calls Intelligent Bias Control. It allows, at lower powers, the SoC to completely shut off the P-cores and run games exclusively on the E-cores. It’s a smart move, as most games are bound by GPU performance, and running high-power P-cores at all times leads to big spikes in frame time and far worse battery life. IBC, as Intel calls it, should offer both better battery life and more consistent performance at lower operating power.

The performance is already looking great, as well. We saw Forza Horizon 6 running on the Acer Predator Atlas 8 at native 1920 x 1200 with the High preset at above 60 fps. Again, that’s without any upscaling or frame generation. We saw similar performance in Clair Obscur: Expedition 33, which is still a fairly demanding Unreal Engine 5 game. We’ll have to evaluate the Arc G3 range in different handhelds and gaming scenarios to see how it holds up, but from our first impressions (and Intel’s bold 42% uplift claim versus the Ryzen Z2 Extreme), the range is shaping up nicely. — Jake Roach

Read more: Intel challenges AMD’s handheld dominance with new Arc G3 chips

Nvidia RTX Spark Superchip

Nvidia RTX Spark Superchip

It’s a rare trade show where a company flexes its all of its silicon and software muscles at once and tries to shift the entire direction of personal computing, but that’s what Nvidia did this week with the RTX Spark Superchip and its broad roster of hardware and software partners as it drives the industry toward its vision for an agentic AI PC future.

It’s one thing to ship a new chip and take an “if you build it, they will come” approach. From what we’ve seen this week, Nvidia has already built the whole damn hardware and software stack necessary to make RTX Spark PCs useful, both as local AI powerhouses and as high-end laptops that run the Windows experience and software that so many already know and love. That’s a monumental task, but Nvidia has already proven that it can execute multiple historic feats of silicon and software engineering in the data center, all at once and at a breakneck pace. Why not try to reinvent the PC while you’re at it?

From close cooperation with Microsoft to integrate its OpenShell agent harness directly into Windows and to improve the Prism x86 emulator, to its work with Adobe and other ISVs to ensure software compatibility, accelerate their applications on the RTX Spark, and ready them for AI agents, to its work with anti-cheat providers to ensure that all of the most popular games run on the Spark from day one, it’s clear that Nvidia is dead serious about making RTX Spark a first-class PC platform.

Six laptop and desktop partners are already on board with Nvidia’s vision, and with a multi-generational silicon roadmap already laid out for future Spark chips, this platform is likely to be with us for many years to come.

We still need to get our hands on these systems to see whether life in the agentic future is everything it’s cracked up to be, but the fall launch for these systems isn’t far away. That the RTX Spark platform is already as mature, broadly adopted, and complete as it is at Computex is a remarkable achievement – even if we just get some really nice Windows on Arm PCs out of the deal. – Jeff Kampman

Read more: Nvidia unveils RTX Spark Superchip for laptops and desktop PCs at Computex 2026

Noctua NL-LC1 liquid cooler

Noctua NL-LC1 liquid cooler

All-in-one liquid coolers and quiet computing don’t traditionally play well together. The small, high-RPM pump that drives liquid through the cooling loop tends to have a prominent sound signature, even at idle, and it always has to be running. But if anybody was going to tame this problem, it’s the engineers at Noctua, which has already made its name by selecting stubborn PC cooling problems and then running them down over years of refinement, testing, measurement, and iteration.

The NL-LC1 cooler embodies every bit of the obsessiveness we’ve come to expect from Noctua. The pump cover assembly is meant to behave as a tuned mass damper that smooths out potential roughness from the frequency components of the pump’s sound, and a multi-layer foam stack underneath the outer shell helps to absorb some of the acoustic energy it produces for lower dBA levels. Two different pump speed profiles and a fully manual mode let PC builders choose their preferred balance between noise levels and performance.

And because it’s built on Asetek’s premium Emma Gen 8 v2 pump design and uses cutting-edge NF-A14x25 G2 and NF-A12x25 G2 fans on its 240mm, 360mm, or 420mm radiators, the baseline performance and sound signature of the NL-LC1 should be strong all around, and it might finally be the liquid cooler that quiet computing obsessives can live with. – Jeff Kampman

Read more: Noctua shows off improved thermosiphon prototype

Dell XPS 13

Dell XPS 13

The Dell XPS 13, once a flagship, top-of-the-line notebook, is back in a new capacity as a champion of the affordable laptop renaissance and a competitor to Apple's MacBook Neo. The system will start at $699, though a temporary sale during the back-to-school season will bring it down to $599.

In some areas, Dell has exceeded the Neo's specs. Its 13.4-inch, 2560 x 1600 screen is larger, higher-resolution, and comes with touch capabilities and a variable 30 - 120 Hz variable refresh rate. The XPS, at 2.2 pounds, is half a pound lighter than Apple's laptop. And Dell has faster USB ports. That's all while keeping the aluminum chassis. We'll still need to test the system to see how its "Wildcat Lake" Intel Core 5 320 processor performs.

Dell is scaling the XPS 13 up, with options for a Core Ultra processor, up to 1TB of storage and up to 32GB of memory, which will surely push prices far higher. But it's the cheapest models that are the most interesting, where Dell may help to reinvigorate interest in Windows laptops under $1,000 with something that looks more like a flagship. — Andrew E. Freedman

Read more: Dell XPS 13 targets MacBook Neo with Intel's Wildcat Lake

MSI Claw 8 EX AI+

MSI Claw 8 EX AI+

Intel took a few tries, but it finally has silicon for a gaming handheld that we're excited about (see above).

While the Intel Arc G3 Extreme processor with B390 iGPU gave us a good first impression, we were also wowed by how it felt to hold, with comfortable grips that are easy to hold and good at distributing the system's 1.7-pound weight. This premium system might just be the one to break out.

The 8-inch, 1920 x 1200 touchscreen has a variable refresh rate up to 120 HZ, and MSI claims it can reach 500 nits of brightness. That's not an OLED, but it looked nice enough in person. The whole system is powered by an 80 WHr battery, matching the Asus Xbox ROG Ally X. Let's just hope it's not too expensive. — Andrew E. Freedman

Read more: MSI Claw 8 EX AI+ brings Intel Arc G3 Extreme to handhelds

Phison E37T

Phison E37T

Crushing NAND shortages have sent prices to stratospheric levels with no end in sight, so any new products that can help bring down SSD prices are a win in our book. Enter Phison’s new E37T SSD controller. This DRAM-less controller reduces the cost of SSDs by stripping out the costly memory required for high-performance models, but through solid engineering and firmware magic, it delivers nearly the same performance as the company’s flagship controller, all while reducing power consumption by nearly half, setting the stage for a new wave of power-efficient SSDs that will run cool without massive heatsinks.

The firm demoed its new silicon running in a laptop, with 14Gb/s of sequential read throughput and 12 GB/s of sequential write performance, along with 3 million random read/write IOPS, all of which comes courtesy of cutting-=edge 4800 MT/s BiCS NAND. The controller also sips a mere 4.5W, a full 2.5W less than its flagship DRAM-equipped E28. Naturally, the DRAM-less design will also have cost savings attached, given that the price of any kind of DRAM is currently apocalyptic. The E37T will begin shipping this year. — Paul Alcorn

Read more: Phison shows PCIe 6.0 X3 SSD controller with 28 GB/s of bandwidth and 6.8 million IOPS, supports 2 petabytes per drive

Asus ROG Strix OLED XG259QWPG Ace

Asus ROG Strix OLED XG259QWPG Ace

If you're a member of the eSports community, Asus has a new gaming monitor for you. Asus claims that the ROG Strix OLED XG259QWPG Ace is the world's first OLED eSports monitor. In terms of response times and color reproduction, OLEDs are on a higher echelon than the typical TN or IPS panels that you'd see in the eSports sector. In this case, Asus is using an LG Display Tandem WOLED panel.

The 540 Hz refresh rate is no slouch either at the monitor's native 1920 x 1080 resolution. Asus says that the 25-inch monitor offers 99.5 percent DCI-P3 coverage, has a typical brightness of 350 nits (1,700 nits peak), and supports VESA DisplayHDR 600 True Black.— Brandon Hill

Read more: Asus' world-first OLED esports monitor can hit 540Hz at 1080p

Asus ROG Rapture GT-BN98 Pro

Asus ROG Rapture GT-BN98 Pro

Asus is among the first manufacturers to show off a production Wi-Fi 8 router, not just teasers of what it will eventually launch. The ROG Rapture GT-BN98 Pro is on track to deliver (or at least come closer to delivering) on the performance promises of Wi-Fi 7.

The ROG Rapture GT-BN98 Pro is a tri-band (2.4 GHz, 5 GHz, 6 GHz) Wi-Fi 8 router that promises up to a 2x increase in median throughput compared to its Wi-Fi 7 counterparts. In addition, IoT devices operating at the outer edges of your router's wireless coverage should also see huge performance gains. The router will also make strides in improving performance and latency in crowded environments.

Aside from the wireless enhancements, the ROG Rapture GT-BN98 Pro also includes two 10 GbE ports along with four 2.5 GbE ports. It also includes a couple of USB-A ports for storage or sharing an internet connection from your phone. — Brandon Hill

Read more: Asus unveils its first Wi-Fi 8 router

Corsair Nightsword v2 Wireless

Corsair Nightsword v2 Wireless

Corsair's Nightsword v2 Wireless SD is a Stream Deck-integrated gaming mouse with a dedicated Stream Deck launch button. It's part of Elgato's Stream Deck ecosystem, which means you can map Stream Deck features directly to the mouse's 8 programmable buttons, or you can use the Stream Deck launch button to open up a "virtual" Stream Deck and activate Stream Deck features from your desktop.

The Nightsword v2 Wireless SD also features Corsair's Marksman S optical sensor, which has a maximum sensitivity of 33,000 DPI and a maximum speed of 750 IPS, and can handle up to 50 G's of force. It sports a right-handed ergonomic design with a sculpted thumb rest and 11 buttons (8 of which are programmable), and weighs 3.14 ounces (89g). Tom's Hardware's Matt Safford had a chance to take a look at it on the show floor, and he said it felt pretty good when he played around with it.

The mouse has 2.4GHz wireless, Bluetooth, and wired connectivity, and gets up to 170 hours of battery life over its 2.4GHz wireless connection with a 1,000 Hz polling rate. It can get up to an 8,000 Hz wireless polling rate, however, but the battery life drops to (a still fairly impressive) 47 hours. — Sarah Jacobsson Purewal

Read more: Corsair shows off gaming mouse with dedicated Stream Deck launch button

Cherry XTRFY K63W Pro

Cherry XTRFY K63W Pro

The Cherry XTRFY K63W Pro is the first 8K gaming keyboard with ultra-wideband technology — a short-range, high-frequency wireless technology that's less prone to signal interference than the more traditional 2.4GHz wireless connection most wireless gaming peripherals use. The keyboard gets true 8,000 Hz polling over both its wired and wireless connections, which means it should be the fastest and most responsive gaming keyboard around when it launches in the U.S. this August.

The K63W Pro also sports a more compact 70-percent layout, which means it keeps the full function row and arrow keys, but drops the navigation row normally seen on 75-percent layouts. This layout is, according to Cherry, "optimized for gaming" as it reduces side bulk to "maximize space for mouse movement." The keyboard also features Cherry's MX Low Profile 2.0 switches and has a gasket mount design for a comfortable typing experience. It will get up to 1,100 hours of battery life over its ultra-wideband connection, "depending on polling rate and RGB intensity" — it's not likely to get this with an 8,000 Hz polling rate, but it's an impressive number nonetheless.

The keyboard launches in the EU in July and hits shelves in the U.S. in August, for €179.99 and $169.99, respectively. — Sarah Jacobsson Purewal

Read more: Cherry XTRFY launches first 8K ultra-wideband gaming keyboard

Gigabyte Aorus X870E Infinity Next

Gigabyte Aorus X870E Infinity Next

Motherboards have, for all intents and purposes, been built the same for what feels like decades. Although the appearance and some hardware features are constantly evolving, the technology used to cool them hasn’t changed much. Slap on a hunk of aluminum and make sure it has enough mass and surface area to do the job. With today’s processors consuming more power than ever, robust power delivery and proper cooling are a must, especially if you plan to overclock. Fast-forward to Computex 2026, where Gigabyte showcased an absolute beast of a motherboard, the Aorus X870E Infinity Next, packed with never-before-used space-grade thermal engineering and a never-before-seen aesthetic.

This E-ATX masterpiece isn’t just your run-of-the-mill Halo motherboard. Gigabyte uses space-grade thermal engineering and rocket-thruster-grade thermal materials, and is the first to use 3D metal printing technology. The VRMs use a 3D metal-printed vapor chamber with an ‘ultra-dense’ omnidirectional Fin Wick and cooling capacity said to exceed 100W. On top of that is an almost alien-like structure that resembles a porous lava rock or even a sponge, which Gigabyte says increases surface area by 44%. Hiding underneath the VRM heatsinks are 16 data-center-class quad OptiMOS MOSFETs (Infineon OptiMOS TDM22544D) for a total of 64 phases (16x4) with a maximum current of 5,120A, nearly doubling the highest value we’ve seen on any consumer-class motherboard. Gigabyte describes it as “limitless power through space-grade and data center-class cooling surface area.” This new tech even makes its way to the primary M.2 heatsink, an AI-Gyroid-designed heatsink with 44% greater cooling area.

Not only does it use cutting-edge materials and processes, but it is, in my opinion, one of the most unique and best-looking motherboards around, with space-age materials on all the parts that get hot (and then some). A full list of specifications isn’t available at this time, but we can glean that it supports a two-slot memory configuration for high-capacity, high-speed CQ-DIMMs, five M.2 sockets, three PCIe slots, and a whole lot more. Availability was not discussed at the Giga event (though they did say it would be produced), nor was pricing mentioned. You can bet your bottom dollar this will not be cheap at all, given all the space-age technology used. It may not be practical for many. It will be affordable for even fewer. But regardless of those details, this is one of the most unique motherboards I’ve ever seen, possibly ever, pushing beyond traditional limits and earning my nod for Best Motherboard of Computex 2026. — Joe Shields

Several AMD partners are showing off the company’s next-generation Helios rack-scale solution running AMD’s EPYC ‘Venice’ processors and Instinct MI455X AI accelerators at Computex 2026 in Taipei, Taiwan. The units are set to become available later this year. There is one major catch, though: they all use UALink-over-Ethernet scale-up connectivity, which may limit their performance in certain workloads that depend on the connection performance. That said, Helios systems with ‘true’ UALink interconnects will also be available.

AMD’s Helios is the company's first rack-scale AI system, and is set to rival Nvidia’s NVL72 VR200 machines based on the next-generation Vera Rubin platform. Helios will rely on AMD’s 6th Generation EPYC Venice CPUs with up to 256 cores, pack 72 Instinct MI455X accelerators with a total of 31 TB of HBM4 memory, and 1400 TB/s of bandwidth. AMD estimates that its performance will be around 2900 FP4 dense PFLOPS, which puts the unit behind Nvidia's VR200 NVL72 system in terms of compute performance, but ahead of it with HBM4 memory capacity. This promises to provide Helios-based systems an advantage in memory-intensive workloads, such as when running large LLMs.

The AI accelerators are interconnected and make use of a UALink-over-Ethernet connection, which provides up to 260 TB/s aggregated scale-up bandwidth (in line with Nvidia’s NVL72 VR200). Helios will also feature Pensando Vulcano network interface cards (NICs), which are among the industry's first 800 GbE network cards that comply with the Ultra Ethernet specification and provide up to 43 TB/s of scale-out bandwidth.

However, the interconnection used on these Helios systems will vary. The machine supports both UALink and UALink-over-Ethernet, but the initial versions will use the latter, rather than the former. This is likely because UALink switches aren't finalized and are pending validation and qualification by AMD’s AI customers.

The biggest advantage of using UALink over Ethernet is that AMD can build Helios using an existing, widely supported ecosystem of validated and qualified components. Ethernet switching ASICs, cables, and other ingredients are already used by hyperscalers and cloud providers worldwide, which accelerates deployment.

But there is a major downside with using Ethernet, even with the UALink protocol on top: Ethernet was originally designed as a general-purpose networking technology; it was never designed to scale up AI accelerators.

As a result, communications may involve higher latency, more protocol overhead, and less deterministic performance than a dedicated scale-up fabric. For large AI training jobs that need all 72 Instinct MI455X accelerators to work in concert, communication efficiency is as important as compute performance. If the UALink-over-Ethernet interconnect cannot keep GPUs fed with data efficiently, some of the theoretical performance of the hardware may be lost in real-world deployments, even though on paper, Helios with UALink-over-Ethernet is as good as Nvidia’s NVL72 VR200 in scale-up bandwidth.

This begs the question of whether UALink will ever be widely used with Helios and whether UALink will ever be widely deployed using copper. Hyperscalers and other companies deploying high-end AI hardware at scale rarely upgrade their hardware.

While the Instinct MI455X certainly promises to be among the best hardware accelerators this year, Helios will likely only be rivalled by Nvidia’s NVL72 VR200. It will be outdated next year when AMD launches its Instinct MI500-series products. These units will be used in the company’s next-generation rack-scale offering, which promises to pack more AI GPUs, potentially requiring optical interconnects with UALink on top. As a result, Helios systems with true UALink interconnections over copper will be on the market for less than a year before those next-generation rack-scale solutions will hit the market.

Of course, nothing is stopping AMD from offering Helios with Instinct MI500-series accelerators and UALink interconnects over copper; however, the company hasn't confirmed the existence of such systems.

The PC industry is bracing for a slip back to DDR4 memory among enthusiasts. Tom’s Hardware has learned at Computex 2026 that both motherboard brands and many module houses — the companies that produce the DIMMs you can buy — are shifting their strategy toward a resurgence in DDR4 platforms as unprecedented memory shortages and price increases continue to raise the entry point into building a PC. Those pain points are particularly acute with DDR5 memory.

We confirmed with over half a dozen sources: motherboard manufacturers and module houses are seeing an increase in demand for DDR4 platforms and shifting production accordingly. This could create challenges, however, as high-performance DDR4 dies, such as the famous Samsung B-dies, are no longer in production. As such, most of the revamped DDR4 kits will top out at a rather pedestrian DDR4-3600.

On the motherboard side of things, at least two vendors confirmed to Tom’s Hardware that they are ramping production of DDR4-supporting motherboards for the second half of the year and into 2027, which makes sense, given that motherboard sales have seen a “collapse” this year, with sales declining by as much as 37% with some vendors. Others have confirmed that they plan to either refresh or re-release DDR4-supporting options later in the year. Many of these products had been in end-of-life (EOL) status, so production lines had long ago shifted to other products. Now new manufacturing capacity will be dedicated to restoring those product families.

This comes as demand for DDR4 platforms has increased. One motherboard brand cited a double-digit increase in sales over the last quarter, which Tom’s Hardware was unable to corroborate, while others simply said the demand has increased significantly.

AMD and Intel have geared up for a shift back to DDR4, as well. AMD launched the Ryzen 7 5800X3D 10th Anniversary Edition at Computex, which the company says it will continue to sell as long as it makes sense; in other words, it’s not a limited edition run. AMD has also shifted the hybrid bonding process of the 5800X3D, setting it up for a more long-term production run. It joins AMD’s Zen 3 XT chips, which it released in 2024 and continues to sell.

Intel continues to sell its Raptor Lake and Raptor Lake Refresh CPUs, as well, and the company told Tom’s Hardware it “will continue to make sure that there are products which can take care of older memory technologies.” One motherboard vendor said it was specifically ramping production of LGA 1700 motherboards with DDR4 support, as options have slowly dried up in the market.

Although there is already DDR4 in the market, it’s also easier to produce, which would help elevate some of the bottlenecks in the current memory supply chain. One of the key shortages right now is advanced packaging, which DDR5 requires with an integrated PMIC. DDR4, by comparison, is much simpler to package and sell, which should help keep prices from climbing into the DDR5 range.

The major chokepoint for DDR4 is wafer allocation, which is a bottleneck that one memory manufacturer (not a module house) pointed to as a potential issue with a shift back to DDR4 platforms. Wafer allocation is a broader bottleneck within the PC industry, however, with Intel even shifting allocation toward the data center as an unprecedented demand for data center CPUs takes hold.

Unfortunately, there appears to be no end in sight. If there’s one thing that nearly all of our sources agreed on, it is that DRAM and NAND shortages will continue throughout all of 2027.

Frore Systems is showing off its LiquidJet Nexus at Computex 2026 in Taipei, Taiwan. The LiquidJet Nexus is a monolithic water block with innovative coldplates, designed to cool two Blackwell GPUs and a Grace CPU, that can replace the complex water block used today. Based on tests conducted by an ODM, the LiquidJet Nexus outperforms the default cooling solution used today and reduces GPU temperatures by around 6ºC, which increases token generation by 10%. Frore intends to build LiquidJet Nexus for Nvidia’s Rubin platforms and is ready to produce them for other accelerators, too.

Earlier this year, Frore introduced its LiquidJet, a coldplate for AI accelerators, tailor-made for exact models of processors using tools originally meant to make semiconductors, in a bid to maximize cooling performance. The LiquidJet Nexus water block builds upon these principles, integrating them into a monolithic unit designed to cool down both GPUs and a CPU inside a server tray. For now, Frore is demonstrating LiquidJet Nexus for Nvidia’s Grace Blackwell superchip, though a version compatible with Vera Rubin is also incoming, we're told.

Frore’s LiquidJet coldplates are made using tools designed to produce semiconductors — using etching and bonding steps — and are architected in accordance with actual thermal maps of CPUs and GPUs they are meant to cool. As a result, they remove heat precisely from hotspots of these processors, and therefore enable better cooling performance than coldplates made using traditional milling methods. Based on tests conducted by a major ODM, Frore’s LiquidJet Nexus reduces the temperature of Blackwell GPUs by 6ºC compared to default cooling solutions, which in turn increases their token generation by 10%.

While a 6ºC temperature drop and a 10% performance improvement may sound humble, these performance improvements impact billion-dollar deployments, where such improvements could mean hundreds of millions in savings. Also, since Frore’s Liquid Jet Nexus is monolithic, it is less prone to leakage. This means less downtime and fewer damaged servers, which means more profits and fewer losses for their owners. Again, since we are talking about billion-dollar deployments, there are significant amounts of money at stake when it comes to cooling and efficiency.

Frore says it is working with the majority of hyperscalers to build LiquidJet-based cooling solutions for their custom hardware. Since LiquidJet is designed to remove 400W – 600W of thermal energy per square centimeter, it can cool down very hot components. Furthermore, since these components tend to scale horizontally, it is not a problem to scale LiquidJet’s performance by increasing its dimensions.

In addition to being more performant and potentially significantly more reliable than existing liquid-cooling solutions for Nvidia Blackwell, Frore’s LiquidJet Nexus also weighs 65% less than rivals and is twice as thin (17 mm vs 34 mm), according to Frore.

While this may not be a significant advantage today (unless you ship your servers by plane), this will be a dramatic advantage for Nvidia’s next-generation Kyber chassis that places servers on their edge rather than horizontally, which will make the importance of LiquidJet Nexus’s weight a bigger factor, as the cooler must adhere to the cooling surface of the integrated heat spreader thoroughly. Meanwhile, it is hard to adhere a massive cooler to a vertically standing motherboard, so one with a lower weight should be easier to attach to the motherboard and chassis without worrying about longer-term deformations.

Speaking of Nvidia’s Kyber chassis, it is worth noting that they are designed for the Vera Rubin Ultra platform, which ups the TDP of GPUs all the way to around 3kW per unit, making its cooling a challenge. Meanwhile, Rubin Ultra GPU scales horizontally by employing a quad-chiplet design, so Frore can address its TDP by reinventing its coldplate, which is easy assuming the company is provided a thermal map of the unit. The same method can be applied to other processors, which is why Frore is indeed working with hyperscalers with custom silicon, in addition to other merchant silicon providers, aside from Nvidia.

Update: After publication, Microsoft provided us with the following statement: “Microsoft Surface Laptop Ultra is not targeting a 110w sustained TDP. Final power will be disclosed closer to availability.” [Emphasis ours]. The original article continues below.

The consumer tech industry is still absorbing the aftershocks of Nvidia's seismic RTX Spark announcements this week at Computex 2026, and there are still many questions around the platform regarding performance, power, and battery life.

One of those questions is the thermal design power (TDP) of the RTX Spark Superchip that's powering the high-end laptops revealed this week. That power budget is everything in a thermally constrained chassis that has to dynamically share power between the CPU and GPU.

A higher power budget in such a system generally translates into higher performance (albeit not necessarily linearly). And if you know the power budget of one platform, you can better reason about its performance compared to other chips with a similar TDP.

Tom's Hardware's Paul Alcorn and I attended a series of lightning-round hands-on sessions with Nvidia's core laptop partners this week, and among other standard questions, we asked representatives from those companies what the power and thermal budgets of their systems were. Unsurprisingly, those partners generally declined to answer.

But Microsoft's reps freely shared that the Surface Laptop Ultra, at least, is designed around a TDP of 110 W for the RTX Spark Superchip.

That figure makes sense, given our experience with the compact DGX Spark mini-PC. That system has an SoC TDP of 140 W, so it's no huge surprise that even the relatively large and well-ventilated Surface Laptop Ultra is designed to dissipate around 80% of that power at peak load.

As a laptop, the Surface also has other components it needs to power, including a screen and any peripherals connected to its USB ports, so extra headroom is required for that purpose. (Other OEMs did disclose that they were including 140W chargers with their devices, so consider that an interesting data point to this end.)

In any event, the enterprising reader might be tempted to extrapolate from our past DGX Spark performance testing and conclude that 20% lower power equals 20% lower performance. But we'd caution against that reasoning.

Chip power and delivered performance generally have a nonlinear relationship past a certain point, and we have no idea what the behavior of the voltage-and-frequency-scaling curve is for the RTX Spark Superchip in the range we're discussing.

We also don't know the full power and thermal management behavior of the RTX Spark Superchip. Modern laptop SoCs (and all chips, really) opportunistically take full advantage of all available thermal and power headroom at the start of a task, so they'll boost up close to the limits of their TDPs while the system is cool before dialing back clock speeds and power to achieve a steady state that avoids overwhelming the system's heatsink and fans. It's not clear how quickly and to what extent the RTX Spark has to clock down to operate within those limits for long-running tasks.

And in a mobile device that has to share power between the CPU and GPU, performance is also going to be highly dependent on the character of the workload. A game, for example, is going to heavily stress the GPU but might not fully occupy the CPU at the same time, while a highly parallel CPU-dependent task like code compilation might fully load the CPU cores without involving the GPU much at all. If you have a (rare) workload that loads down both of those functional units at once, overall performance is likely to fall further than with one that only demands one type of processing resource or the other.

It's also worth remembering that every laptop is different, and power envelopes are carefully tuned for each chassis to best balance design constraints between SoC temperatures, skin temperatures, and noise, among other factors.

All that said, from what we've seen, Microsoft's 110W target seems likely to be typical of the 15"-16" laptops that other OEMs plan to introduce. Logically, it's also likely that we might see lower power budgets for thinner or smaller systems.

But the short version of all this is that there's a lot we still don't know about the RTX Spark platform, and we're still a long way from the launch of the laptops with this chip inside. We expect to learn more about this platform, its design targets, and its behavior in the coming months as we lead up to Nvidia's projected fall launch. Stay tuned.

With the show floor officially open, there's plenty to see and do at the Nanggang Exhibition Center at Computex 2026. However, the reality is that for most of our team, that's not the only place that they will have to visit. At events like these, companies regularly schedule meetings outside of Computex 2026 itself, either to secure more space for their products or to show off products and concepts that aren't strictly ready for the show floor and the thousands of attendees.

In our Day 3 blog, our team of staffers has crossed Taipei, photographed exciting new hardware, and experienced the dizzying halls of TaiNEX 1 and 2. If you've not caught up on their journeys so far, be sure to read Day 0, Day 1, and Day 2 first.

Paul Alcorn: Editor-in-Chief

Another blistering day in the heat of Taiwan started with a meeting with a high-ranking Micron executive to talk about the state of the memory and storage industry. I learned a lot that I will crystallize into an article soon.

After that nice air-conditioned chat, I hit the halls, moving between multiple vendors to see the latest the industry has to offer. Asus, ROG, MSI, Patriot, Adata, and many others kept me busy throughout the day, particularly as I dug deeper into the side effects of the memory shortages. Unfortunately, not a single individual we spoke with expects any sort of recovery soon, and this will have dramatic effects on smaller module makers, the companies that make SSDs and DRAM, and a cascading effect on the other OEMs.

In fact, though Computex easily had record attendance, the halls were busier on day two than I have ever seen in my 15 years of covering this event; there were relatively few substantive announcements on the PC front. In fact, even the announcements on the data center and ODM side were exceptionally light, largely due to the impacts of supply shortages and the resulting uncertainty that has slammed the brakes on new product development.

Joe Shields: Staff Writer, Components

Three days in, and another day of 90-degree heat. It’s amazing how you can break a sweat here just by walking outside early in the morning. Today’s adventures included floating back and forth between the halls and meeting with several companies. I’ll tell you, if I see another booth with CDUs (Cooling Distribution Units), copper water blocks for AI servers, or PDUs (Power Distribution Units) supporting the AI Data Center boom, I’ll explode.

The highlight of the day has to be my second meal, Mushroom Risotto and Steak. But, because I was so entrenched in work, recovering from a spate of unexpected poor photography on my part that nearly brought my work, well, one article, to a standstill. Afterwards, I just went to bed, physically drained and mentally wiped out. Sleep… Now that’s an escapade.

There’s nothing like crashing when you’re mentally thrashed and waking up a couple of hours later practically wide awake—damn this time change. Tomorrow I’ll be all over the place, with my first meeting at a hotel, two back at the convention center, back to the same hotel for another meeting, and finally back to the convention center again. It’s certainly not the most efficient logistics, but that’s how the cookie crumbles at events like this. The most interesting man in the world says, “Stay thirsty, my friends,” but I’ll say, “Stay hydrated, friends.” One more day to go.

Jake Roach: Senior Analyst, CPUs

Well, I had to open my schedule to find out what I did today in order to write this, so if that’s not a testament to where I’m at in the Computex arc, I don’t know what is. I started my day with my final AMD roundtable of the week, this time focused on ROCm, before taking the MRT down a few stops to Intel’s demo showcase. I met with Dell and Samsung Display in the afternoon, taking a look at the XPS 13 and Samsung Display’s new QD-OLED panels, but much of my day was spent running around to various vendors, chasing down a few key stories (stay tuned on that front).

There are a lot of products at Computex, and it’s impossible to cover them all. But the most interesting conversations happen around the products, not explicitly about them. Although my gung-ho attitude earlier in the week is starting to catch up with me, I’m hitting a sleepy second wind as we close out Computex and start to peel back the curtain on what the broader industry thinks about where things are headed. Also, man, it was hot today.

Jeffrey Kampman: Senior Analyst, Graphics

I began this witheringly hot Wednesday with a surprise trip to the TICC, where Nvidia has set up its enterprise demos for this year, for some quick conversations with the leaders who are guiding the rollout of Vera Rubin and paving the way for the agentic AI future in the data center.

From there, I hopped onto the blessedly air-conditioned MRT and headed back to the Nangang Exhibition Center to criss-cross the show floor and visit some of the biggest booths at the show, including Asus, Asus ROG, MSI, and Gigabyte. And for as tough as it is out there in the consumer PC space, it’s still hard not to feel a sense of awe and wonder when you’re standing in the very center of TaiNEX 1, surrounded by countless thousands of tech enthusiasts and industry professionals, all of them hyped up by the dizzying array of new and sometimes wild stuff on display.

I closed out the day by visiting the fine folks at Noctua with Editor-in-Chief Paul, where we got a thorough deep dive into the company’s obsessively engineered solutions to problems that many PC builders wouldn’t even consider problems. Would any other company contemplate placing a tiny tuned mass damper on top of an all-in-one liquid cooler pump to smooth out its noise signature? Probably not, but Noctua did, and my ears are sensitive enough to this kind of thing for me to be seriously interested. For now, though, I am desperately in need of rest.

TeamGroup showed Tom's Hardware a new external SSD with a remote “self-destruct” function at Computex 2026. The T-Create Expert P35SG External SSD offers a “wireless one-click data destruction” function, letting you wipe the contents of the storage device remotely with a single text message.

The drive is connected wirelessly via 4G LTE to ensure that the user can still reach it even when not connected to Wi-Fi or to another device. Once it receives the wipe command, it triggers a “deep-level logical wipe” of the drive and then executes a high-voltage “physical hardware breakdown” to ensure that its contents are unrecoverable. This new drive arrived on the show floor nearly a year after TeamGroup revealed the P250Q-M80 M.2 SSD that had a similar physical self-destruct function.

Aside from this wireless high-security external drive, the company also unveiled the T-Create Expert P35S, which offers the same “self-destruct” function, but through a physical button right on the drive. There’s also the P33 External SSD, which includes an e-Paper display that can show important information like available storage capacity, drive health, drive name, and other information. This lets you see what a particular USB drive contains without needing to plug it into a device or place an external label on it.

There’s also the T-Create Expert AI 4R CUDIMM — this quad-rank CUDIMM packs in two rows of DRAM chips on both sides, giving a single RAM stick 128GB capacity. The sample DDR5 RAM was installed on an MSI MEG Z890 Unify-X that offers two RAM slots, giving it a total capacity of 256GB. We’ve seen other vendors like Origin Code bring 4R CUDIMMs to market, too, and this high-capacity RAM, which could potentially go as high as 256GB per stick, is quite useful for AI developers and enthusiasts who need the massive amounts of memory for AI agents.

Gamers gets goodies from T-Force too

Aside from the professional and creator-focused T-Create product line, TeamGroup also introduced new RAM kits, SSDs, and associated accessories from T-Force. The most interesting new device is the T-Force Liquid II SSD cooler, which brings liquid cooling to SSDs. This cooling solution is designed to sit on an exposed SSD and uses aluminum alloy heatsink fins, a micro cooling fan, and a liquid coolant (available in multiple colors) to help keep your high-performance storage drive cool. The coolant itself is designed to last five years and could be topped up as necessary.

There’s also the T-Force DARK RGB desktop memory kits, which are designed for low-profile builds with a maximum height of just 42mm. Despite its lower height, this RAM kit still features 6400 MT/s speed and a full RGB light strip. It’s also available in various capacities up to 64GB, making it an ideal solution for enthusiasts looking for an SFF build.

And, in celebration of 10 years of T-Force, TeamGroup also released several carbon-fiber-themed products. These include the T-Force Delta RGB DDR5, T-Force Xtreem DDR5, T-Force Z54E M.2 PCIe 5.0 SSD, M400 RGB External SSD, and the T-Force Vulcan and Delta LPCAMM2 DDR5 memory.

We expect these products to drop in the coming weeks and months, although we weren't able to confirm exact availability dates. Nonetheless, we can't wait to get our hands on some of these new solutions and see how they match up against the best RAM and SSDs we have available at the moment.

MSI has long been a partner of Intel's with its Claw handheld, but at Computex, the company's Claw 8 EX AI+ looked like the strongest version yet. That's largely based on the fact that it's using the Intel Arc G3 Extreme, based on the company's Core Ultra Series 3 (Panther Lake) chips, with integrated graphics ready to take on AMD.

The system, which comes in one color, 'void purple,' has two extended grips on the sides. On the ground at Computex in Taipei, Tom's Hardware's Jeffrey Kampman confirmed that the Claw 8 EX AI+ "felt excellent in my hands," and that "the grips are large and well-contoured and distribute the weight of the system well."

Those grips also feature hall-effect triggers and sticks, and what MSI refers to as a "highly responsive D-Pad," along with a linear motor for rumble feedback.

Ergonomics have become a big deal. When Asus moved to its ROG Xbox lineup, it added some pretty serious grips to the sides. It makes the system wider, but also far more comfortable to hold. The same holds true here.

The system boasts an 8-inch, 120 Hz display with a variable refresh rate. The screen isn't totally flush between the grips, and juts out below them a bit in a weird style choice.

In our hands-on time with the device, Hogwarts Legacy ran well on the integrated B390, but we'll need to spend time testing in a less controlled environment to learn more for sure.

While MSI is still using its own Control Center M, its integrated into the Xbox Full Screen Experience, so you shouldn't have to deal with the Windows desktop unless you truly want to.We're hoping that the 80 WHr battery (the same size that you find in the Asus ROG Ally X), will allow for at least a few hours of gaming on a charge. Panther Lake has proven to be fairly efficient, but gaming is a pretty heavy workload.

Variations of the Claw will come with up to 32GB of RAM. MSI hasn't listed storage, but points out there's a single M.2 2280 SSD slot on the Claw. While MSI has yet to announce specifics on release dates or pricing, an early page on Best Buy that has since been pulled down suggested that a version with 32GB of RAM and 1TB of storage may be $1,699.99. But like much at Computex, there's not official pricing, so it's all still up in the air.

Beyond using GaN tech to shrink their size and including ever-higher wattage to meet the demands of high-end GPUs and local AI, PC power supplies aren't exactly a space known for flashy innovation. But Lian Li is having none of that.

The company's upcoming Platinum V2 line, expected to arrive in September, keeps the extended lip with 90-degree-angled connections for ease of installation and less cable mess, while adding a few more new tricks.

An LED on the side of the unit switches from green to red when it's time to clean the supply's intake fan. And much like higher-end cases, there is a magnetically attached filter over the fan that you can pop off for convenient cleaning.

If that weren't enough, the inner edge of the PSU houses an expansive four-pin fan (6 connectors) and USB 2.0 (4 connectors) hub, which snaps on via magnets and pins, so you can remove it to connect all those fiddle wires, then pop it back on the PSU. The hub itself connects to a USB 3 header, which includes a passthrough so you don't have to sacrifice what might be the only 20-pin header on your motherboard.

My main concern is that plugging all your fan and accessory cables right next to your power cables could result in exactly the kind of cable jumble that the Edge line was designed to help with. And shoving that many cables next to each other in a tight space doesn’t sound great if something goes thermally sideways. But we'd have to try one out to say for sure how much of an issue that might be.

Lian Li hasn't divulged pricing yet, but it does plan on launching 1,000, 1,200, and 1,350-watt models of the Edge V2 in September.

Tom's Hardware stopped by the ASRock booth, and we were greeted by a wide variety of products at this year's Computex. From AI machines and new AIOs to power supplies and a one-click OpenClaw installation, there was a lot to take in, but not much new on the component side beyond the new AIOs and power supplies. The 10th-anniversary edition Taichi products they showed off had their own unique look, but unfortunately for us, they were concept models and will likely not be released.

Starting with the AIOs, the new Taichi Aqua 360 LCD cooler is a hybrid of a custom loop and an AIO of sorts, with features like an inline flow meter, G 1/4 connectors for expansion, a dual-pump architecture said to support a 500W TDP (and for redundancy). The radiator is thick at 38mm and comes with an ‘aerospace grade’ LCP fan. The pump/block combo uses the latest-generation Asetek pump to push water through the loop running up to 4,000 RPM. It also has a unique dual-mode Top cover design with a detachable magnetic 3.4-inch LCD that shows system information. When removed, it reveals the transparent design, showing the water flowing in the loop and lit by RGB LCDs. Pricing wasn’t mentioned, but we should see this available around Q3 2026.

The Taichi 360 Holo’s claim to fame is the implementation of a unique 4-inch holographic feature on top of the pump to display images or system information. Its POV-based (Persistence Of Vision effect) holographic film spins like a fan, creating a floating, customizable, 3D-effect image that’s truly unique to AIOs. Like the Taichi Aqua, it’s only found in a 360mm size, but it is a bit thinner than the Aqua 360.

Moving down the product stack, ASRock showed off its new Rock and Challenger series AIOs that will be much cheaper than the flagships above. The Rock 360 Digital sports a small digital display on the pump/block, while the Rock (non digital) forgoes the display and instead has a small, frosted RGB line on top of the block. The showstopper on these is an awesome price point. The Rock 360 digital comes in at a low $69.99, with the plain Rock at a wallet-sparing $59.99.

ASRock is diving headfirst into power supplies, expanding its portfolio into the AI realm with the Taichi WS (Workstation) line, which consists of three models ranging from a mere 2,600W to a monster 3,200W flagship capable of supporting up to FOUR RTX 5090s. They feature ASRock’s cable Over-temperature protection, which, as the name implies, shuts the system down when it detects that a GPU's 12-V2x6 connector has overheated.

We also saw the Phantom Gaming lineup, which is 80Plus Platinum-rated and earned a Cybenetics A++ rating (850W model; 1KW is A+) for noise, so it should be not only efficient but quiet, too. These come in 850W and 1,000W models and also use Cable Over-temperature protection. They also updated the Steel Legend SKUs (80Plus platinum) and the Pro models for entry-level SKUs. Pricing or availability was mentioned on these.

Samsung displayed its first physical mockup of HBM5 memory at Computex 2026 in Taipei. Tom's Hardware dropped by to see its pairing of the eighth-generation AI memory with a new in-package cooling structure it calls Heat Path Block, or HPB. Just last week, rival SK hynix unveiled its own iHBM thermal design, meaning both companies are now focusing on the same heat bottleneck in the die-to-die interface that connects memory to the processor. Samsung also confirmed that it’ll fab HBM5’s base die on its in-house 2nm process, down from the 4nm node used for HBM4 and HBM4E.

Rather than letting heat escape outward through the core dies, HPB builds a separate set of thermal pillars that pull heat from inside the stack and carry it to a spreader sitting above or beside the package, according to Samsung at Computex.

The design concentrates on the D2D PHY layer, the high-speed link between the HBM base die and the GPU, where power density and temperatures increase exponentially as stacks grow taller and run quicker. Samsung said it has already implemented and verified HPB on HBM4E, the generation whose first 12-layer samples it began shipping last month at 14 Gbps, scaling to 16 Gbps, with 3.6 TB/s of bandwidth per stack.

Samsung runs both a memory business and a logic foundry, letting it build the HBM5 stack and the 2nm die beneath it in-house. "AI systems are becoming more powerful and densely integrated, making heat management, data-processing efficiency, and packaging stability just as important as memory performance itself," Song Jai-hyuk, president and CTO of Samsung's Device Solutions division, told reporters at Computex, according to the Korea Herald. Song said the company would keep building its competitiveness in next-generation memory through cooperation with partners, including Nvidia.

Last year, a roadmap from KAIST projected HBM5 reaching a 4,096-bit interface, roughly 4 TB/s per stack, and about 100 watts of per-stack power, a thermal load that goes a long way in explaining why both Korean memory giants are reworking their packaging now rather than at launch.

SK hynix faced the same problem via a different route. Its iHBM design embeds cooling elements made of electrically non-conductive, thermally conductive silicon into the D2D PHY layer, which the company said cuts thermal resistance by more than 30% against current products.

SK hynix has opted to place a cooling element directly at the hotspot, while Samsung has built a route to evacuate heat away from it. Both methods are slated to debut with HBM5, but it’ll be a little while before we see either in action, as neither company expects it to reach mass production before 2028.

Computex 2026 is well underway, and we’ve provided you with pre-show coverage and all the official announcements this week. You can catch our Day One coverage here and keep track of our dedicated Computex 2026 hub.

Intel acknowledges missteps with Arrow Lake

There’s no question that Intel failed to hit the mark with enthusiasts when it introduced Arrow Lake. Performance regression in games compared to previous-generation chips was the big downer for Arrow Lake, with Intel later blaming developers for not fully optimizing for the new architecture.

However, Intel is now laying it all on the table, saying that it needs to do better. “From an enthusiast perspective, it was... we needed to build back our reputation. I am sure you would agree with that, and this was, hey, [we’re] making sure we are providing value to the gamers, and we start with Arrow Lake Refresh, and we have a very strong roadmap to come,” Intel told us.

• Intel addresses Arrow Lake blunder: 'We needed to build back our reputation' — says Arrow Lake Refresh's low price a key first step, laying the groundwork for Nova Lake

Wi-Fi 8 is on the horizon, but Wi-Fi 7 is by no means going anywhere

Just when we are getting used to Wi-Fi 7 routers and prices have fallen to reasonable levels, Wi-Fi 8 is just around the corner. At the show, Asus revealed its first Wi-Fi 8 gaming router: the ROG Rapture GT-BN98 Pro. While the new router shares a design with the two-year-old ROG Rapture GT-BE98 Pro, it features a brand-new chipset that promises to offer vastly improved real-world performance over its Wi-Fi 7 predecessor.

And while Wi-Fi 8 was the talk of the show, MSI touted some high-end Wi-Fi 7 routers, including the RadiX BE19000. The tri-band router amazingly includes an internal PCIe SSD slot. This allows the router to serve as a NAS for file sharing and device backups. Also onboard are two 10 GbE ports and four 2.5 GbE ports.

• Asus unveils its first Wi-Fi 8 router
• MSI unveils latest set of WiFi 7 gaming routers touting ultra-fast speeds

Everything Else...

We can’t get into detail on everything we’ve seen so far at Computex in this wrap-up — we’ll let the individual news stories speak for themselves. Here’s everything else we’ve covered for Computex 2026 Day One:

• Cooler Master shows off new MWE Gold V4 Power supplies and GPU Shield adapter
• Microsoft debuts Surface RTX Spark Dev Box — Nvidia-powered mini-PC helps devs get ready for an agentic Windows
• Cherry XTRFY launches first 8K ultra-wideband gaming keyboardCorsair shows off gaming mouse with dedicated Stream Deck launch button
• Noctua's first AIO starts at around $250 for 240mm variant, features silent Asetek pump and NF-A12/A14 fans
• NZXT showcases H6 mid-tower chassis, new Ultra RGB fans, and a white H2 offering

Corsair showed off the latest iteration of its lightweight gaming headset lineup — the Corsair HS35 v3 headset, in both wired and wireless versions, at Computex 2026. These lightweight headsets feature a new floating headband design, 50mm dynamic drivers, detachable boom mics, and Dolby Atmos Spatial Audio compatibility.

Both versions of the HS35 v3 feature custom 50mm Neodymium dynamic drivers with a frequency response of 20 - 20,000 Hz and an impedance of 32 ohms. The wired version weighs a mere 8.11 ounces (230g), which makes it one of the lightest gaming headsets we've seen (though, not the absolute lightest — Logitech's G325 weighs just 7.47oz / 212g) — while the wireless version is slightly heavier at 8.82 ounces (250g). The wired version is over 2.5 ounces lighter than the HyperX Cloud III (10.86oz / 308g), which is one of the comfiest headsets we've tested, while the wireless version is almost 2 ounces lighter than the Turtle Beach Atlas Air (10.61oz / 292.5g), which is, well, basically as light as air. Weight doesn't necessarily guarantee comfort, but it's certainly a major factor.

The headsets also feature a "floating" headband design — an elastic strap connected to both sides of the headband, so the headband doesn't sit directly on your skull and instead "floats" above it — which is a design we've seen several of our favorite comfortable gaming headsets use. Other design features include a lightweight plastic frame, memory foam-padded earpads covered in breathable mesh fabric, and a detachable omnidirectional boom microphone with on-ear controls for both muting and volume control.

The wireless version of the HS35 v3 features 2.4GHz wireless, Bluetooth, and wired connectivity, while the wired version has a universal 3.5mm wired connection. Corsair says you'll get up to 30 hours of battery life over the wireless version's 2.4GHz wireless connection at 50% volume, which isn't the most impressive battery life we've heard of but is probably more than enough for someone who doesn't play days-long gaming marathons.

Both the HS35 v3 wired version and the HS35 v3 Wireless are available now, for $49.99 and $79.99, respectively, in black and white colorways.

Representatives told Tom's Hardware at Computex 2026 that counterfeit memory modules with G.Skill and V-Color badges have hit various marketplaces in China, and the problem is widespread enough that V-Color has observed lower sales to certain clients. Fake memory modules are nothing new, but in this case, forged DIMMs reportedly use printed circuit boards (PCBs) that are identical to those used by G.Skill and V-Color, which makes early detection of counterfeit products difficult.

“We have customers telling us that some Chinese manufacturers are making copies of our products, as well as products from other brands, and selling them at lower prices than we do,” a spokesperson for V-Color told Tom’s Hardware. “We do not really know how they are doing it. The PCB looks the same, even the heat spreader looks identical.”

Given the current situation with memory prices, counterfeit memory modules have become a growing problem in Asian online marketplaces, particularly in gray-market and second-hand channels. Recent reports described fake DDR5 modules containing dummy ‘chips’ made of plastic or fiberglass with fake markings, as well as mislabeled modules marked as products from well-known brands. However, the new case is different because forged memory modules use PCBs that are identical to those used by G.Skill and V-Color and use the same heat spreaders and perhaps complex RGB lighting, making distinguishing between real and counterfeit products particularly difficult.

While developing a PCB for an enthusiast-class memory module takes time and money, producing that PCB is fairly easy, as makers of DIMMs use neither complex PCBs nor too complex equipment. Also, developing enthusiast-grade memory modules (ensuring that they run at the right specifications with popular CPUs) also takes time and money. In addition, sourcing chips from DRAM makers or distributors to make fake memory modules is hard, if even possible at all today. Finally, quality control at the chip level and at the module level is relatively costly, and these two are the main value adders when it comes to enthusiast-grade memory. That said, forging high-end memory modules is easy and might be a profitable business, especially if perpetrators have access to second-hand memory chips. Speaking of chips, the V-Color representative could not say which ICs the fake modules used.

“We do not know because we have not had any of those modules in our hands,” the spokesman said. “We only use SK hynix memory, so we cannot say what chips they are using. We have not received any RMAs related to them. Other brands may have received returns and been able to inspect the products, perhaps because the memory chips were different.”

While a representative for G.Skill confirmed that there are counterfeit G.Skill memory modules sold on Chinese marketplaces, he said that the issue has existed forever, so users should buy their high-end DIMMs from official partners and resellers in a bid to avoid the problem.

Noctua originally announced and showed off its first liquid cooler last year, promising to bring the same performance and aesthetics of its iconic air coolers to AIOs. Today at Computex 2026, the company has officially unveiled the NL-LC1, putting a name on the lineup that features 240mm, 360mm, and 420mm variants. The NL-LC1 will hit store shelves two weeks later, on June 16, starting at 220 Euros (around $250).

This AIO was originally intended to launch in Q1 2026 but was delayed to Q2 late last year. Hence, we already know most of the details about this product, including the available sizes, mounting, pump solution, fan setup, and more. Most importantly, the NL-LC1 uses Asetek's Emma V2 platform for its pump rather than an in-house solution, but Noctua has developed a custom noise-suppression system around it.

There are three layers of insulation meant to ensure quiet operation, and an integrated mass damper called the "NL-PNA1" (Pump Noise Absorber) to reduce vibrations. Noctua is not only limiting the amount of noise but also ensuring the pitch isn't annoying, even for what little you might be able to hear. The company's own charts show a notable difference with and without the top cover that provides this soundproofing.

The cooler comes with Noctua's legendary NF-A12x25 G2 (120mm) fans for the 240mm and 360mm models, while the 420mm features the NF-A14x25 G2 (140mm) fans instead. Similar to Silverstone's IceMyst Pro coolers we reviewed a couple of months ago, there's also an option for an additional auxiliary fan that can channel airflow to motherboard VRMs, RAM, or M.2 SSDs. It's called the "NL-ACF1" and costs 20 EUR ($23).

Noctua has equipped the NL-LC1 with its SecuFirm2+ mounting system, which supports all current-gen and recent CPUs on the market — AM4, AM5, LGA 1700, and LGA 1851. If you're already using a brown and beige air cooler, it should be a simple swap for you without needing to replace the mounting bracket. Lastly, there are three pump profiles corresponding to different fan speeds and noise levels.

By default, the pump runs at a minimum of 750 RPM, but it can be dialed up to 3,400 RPM manually, using the "Manual" profile. The "Quiet" profile limits the speed to 2,100 RPM, while the "Balanced" profile lets it reach 2,600 RPM. The moment the liquid temperature exceeds 30°C, the Balanced profile will dynamically increase the pump speed up to 3,400 RPM at 45°C to maintain a balance between cooling and noise.

Available from June 16, the NL-CL1 will cost 220 EUR for the 240mm version, 250 EUR for the 360mm version, and 280 EUR for the top-end 420mm version. Converted to USD, that's a ~$250 starting point. Today, you can get high-quality Thermalright AIOs for $50, and spending a bit more will get you their fancy LCD options, too. The 420mm variant comes out to $325 — Arctic's Liquid Freezer III Pro 420 costs around $100, and we gave it a positive review earlier this year.

Even if reviews paint the NL-LC1 in a glowing light, that pricing is a tough pill to swallow for anyone but hardcore Noctua fans. Moreover, it's important to mention that this is different from the pumpless AIO that the company has been showing off. That's still in the prototype stage and not intended for mainstream markets, though the company did say that in the "Quiet" profile, the NL-LC1 is beaten by the pumpless model.

With the first few days firmly behind our team on the ground at Computex 2026, Day 2 is the first day that our crew got their boots on the ground on the show floor of the Nanggang Exhibition Center in Taipei, Taiwan. With a busy schedule of meetings, interviews, and presentations in tow, it's been a busy few days, and we're now deep in the throes of covering the event. This series of diaries from our team is intended to give you insights into our thoughts and feelings about covering a massive event like Computex 2026, documenting the trials and tribulations that naturally come with event coverage.

If you've not read them yet, be sure to check out our Day 0 and Day 1 Tom's Hardware Unfiltered blogs to catch up on what they've been up to so far.

Matt Safford: Managing Editor

As I hauled my increasingly complaining body through eight meetings today, I heard a familiar refrain from multiple case and cooler makers: they’re focusing on lower-cost designs (often sub-$100), because they know how cash-strapped gamers and PC builders are, thanks to the ongoing RAM, storage, and GPU pricing crisis. And to be fair, they are trying to sell PC hardware to cash-constrained consumers at perhaps the toughest time for our little hobby/obsession in decades. So, of course, the focus is on lower-cost components and hardware.

Asus, meanwhile, in celebrating the 20th anniversary of its ROG sub-brand, launched an expansive lineup of mostly rehashed Edition 20 products, at some of the highest prices I’ve seen. Take the $599 gaming keyboard, for example, or the $250 mouse. I tested them both, and they’re generally great, with high-end specs and solid performance. But they’re also wildly expensive at a time when key components (not to mention gas and groceries) remain sky-high. It’s easy to call Asus’ Edition 20 launch tone-deaf (did I mention that many of these products feature real gold-plated accents?).

But I can also see why the company (and ROG die-hards) would want to celebrate 20 years of a sub-brand that, let’s face it, has generally offered up high-priced (and generally well-regarded) hardware. But I would love to know how successful the Edition 20 lineup turns out to be, six months or more down the road. Because it’s hard to imagine throngs of people excited to buy a $1,000-plus gaming chair, or the many other high-priced Edition 20 products. But I can certainly imagine lots of people, many of our readers included, looking for a PC component bargain after shelling out for 32GB of RAM or a 2TB SSD at today’s prices. Maybe Asus should start thinking about those kinds of customers, too.

Joe Shields: Staff Writer, Components

Today, Cooler Master HQ was my first stop before heading to the Convention Center and more traditional booth visits. I have to admit, it was a bit overwhelming at first, trying to find my first meeting location and navigating through a sea of people everywhere, but mostly because I left my schedule with all the time and location information back at the hotel (thanks, Jake Roach, for the save!).

It’s very different from CES, where most companies have their own suite or ballroom at one of the Vegas hotels, and it’s a little more private. Outside of that, I had my first taste of 90-plus-degree weather with over 70% humidity. Even walking across the street to the other Hall, I broke a sweat. Oppressive when you’re not used to it. Even though I have almost twice as many meetings tomorrow as today, it doesn’t feel as intimidating after going through it all today, and all but the last meeting are in the same location.

Jake Roach: Senior Analyst, CPUs

I spent yesterday with Intel, so today was spent with the other guys: AMD. I sat in on three roundtables today with AMD’s Rahul Tikoo and David McAfee, discussing everything from the EXPO ULL to AMD’s reaction to the RTX Spark. I also spent some time in an Intel Q&A post-keynote, where I asked the company about its rumored 3D V-Cache competitor, and it gave me the exact answer I expected — “stay tuned.”

Despite being chipper — Paul (Alcorn) even said I looked “peppy” today — I am waiting for my body to crumble in on itself. I am beyond dehydrated. I haven’t been eating anything until the very end of the day, and there are other issues that I will spare you the details of. That’s a problem for future Jake, however. Current Jake is excited to look at chips.

Jeffrey Kampman: Senior Analyst, Graphics

My agenda today was once again dominated by Nvidia. We were back at the Grand HIlai early to attend a press Q&A with an undercaffeinated and punchy Jensen Huang, who was by turns happy to discuss the RTX Spark platform and unhappy with members of the press who failed to ask what he considered to be good questions. After that, we got hands-on time with RTX Spark laptops from five of the six major vendors who will lead the charge when these products come to market this fall, and we picked up some interesting details of those products that have been hard to come by from official channels.

After those briefings, I finally picked up my Computex badge proper and got a bit of floor time at the Nangang Exhibition Center, and I’ll be back tomorrow to visit more booths. The show is truly in full swing now!

Corsair showed off its new Stream Deck-integrated gaming mouse at Computex 2026 — the Nightsword v2 Wireless SD Stream Deck gaming mouse. This mouse is semi-unique in that the Stream Deck app detects it as part of the Stream Deck ecosystem, which means you can map Stream Deck features directly to the mouse's 8 programmable buttons. It also features a dedicated Stream Deck Launch button, which lets you open up a Virtual Stream Deck (e.g. a Stream Deck on your screen) and activate Stream Deck functions from said "VSD."

This means the Nightsword v2 Wireless SD is more than just a mouse with macro-programmable buttons — not only do you get access to the Stream Deck's wide range of integrated plugins (mostly for various streaming and gaming platforms — it is called a Stream Deck for a reason, after all), but you can also program Stream Deck functions such as "multi-action," which lets you set up multiple Stream Deck actions to be performed in a sequence. It's great for people who want a one-click way to access their streaming or gaming setup.

Of course, the Nightsword v2 Wireless SD isn't the first gaming mouse with Stream Deck integration — Corsair's Scimitar Elite Wireless SE also featured native Stream Deck integration that allowed for both Stream Deck programmability and access to Virtual Stream Decks. However, instead of having a dedicated Stream Deck button, the Scimitar Elite Wireless SE just kind of had Stream Deck functionality built-in as a bonus (although, as a mouse with a 12-button side panel, it perhaps had more of an argument for the programmability).

Specs-wise, the Nightsword v2 Wireless SD features 11 buttons (8 of which are programmable) and sports Corsair's Marksman S optical sensor, which has a maximum sensitivity of 33,000 DPI and a maximum speed of 750 IPS, and can handle up to 50 G's of acceleration. This is the same sensor we've seen in Corsair's other recently-launched mice, including the Sabre v2 Pro Wireless CF and Sabre v2 Pro Wireless MG as well as last year's Scimitar Elite SE. It has a sculpted, right-handed ergonomic design that measures 5.1 x 3.07 x 1.61 inches (129.6 x 78 x 40.9mm) and weighs 3.14 ounces (89g). The primary mouse buttons have optical switches, which are rated for up to 100 million clicks.

The mouse features 2.4GHz wireless, Bluetooth, and wired connectivity, and has up to an 8,000 Hz polling rate over its 2.4GHz wireless connection. It gets up to 170 hours of battery life over its 2.4GHz wireless connection (with a 1,000 Hz polling rate and the RGB lighting turned off). If you're using an 8,000 Hz polling rate, that number drops to 47 hours (with the lighting turned off) or 25.5 hours with the lighting turned on — which is better than we expected.

The Corsair Nightsword v2 Wireless SD is available now for $129.99.

AMD announced EXPO Ultra Low Latency at Computex 2026, but provided very few details about what it is or how it works. The company claims it provides a 13% average uplift compared to JEDEC standards, and a 4% uplift compared to standard EXPO, and that’s it’s coming from leading memory partners. Now, AMD’s David McAfee, VP and general manager of Ryzen and Radeon, has given Tom’s Hardware a bit more color on what exactly EXPO ULL is and how it will roll out.

“Expo has evolved, and we've been working with all of the module vendors. We've seen an opportunity with adding more sub timings to the SPD profile, and just getting a little bit lower latency,” McAfee told Tom’s Hardware.

Beyond what EXPO ULL is doing, AMD says that it will work with existing chipsets and motherboards, but McAfee clarified that he wasn’t sure if it would require a BIOS update (that seems likely). Regardless, McAfee said: “My advice to all of your readers and viewers is: update your bios.”

EXPO ULL kits will come with different branding on them, including a new badge, so “it’ll be noticeable that these kits are different.” McAfee also says that, although AMD doesn’t control memory or module prices, “[AMD’s] understanding from the partners is [that] they expect to bring these in at effectively the same price points that the current kits are at.”

Those price points, at the moment, are inflated due to ongoing memory shortages. But McAfee isn’t being coy here about pricing. The executive says ULL is “simply about extending the sub-timings and really getting every little bit of OC performance out of those DIMMs,” so there’s nothing specifically about ULL that would cause the DIMMs to be radically more expensive.

Across a test suite of 30 games with the Ryzen 7 9700X, AMD says ULL delivered a 4% uplift compared to standard EXPO, with both running at DDR5-6000, and a 13% uplift compared to JEDEC-standard DDR5-5600. 1% lows, as expected, see more of an improvement, with a 15% uplift compared to JEDEC and the same 4% compared to standard EXPO.

AMD says that EXPO ULL is “coming soon,” but it hasn’t said anything beyond that for a release window. Regardless, it already has partners lined up to support the feature, including G.Skill, Kingston, Lexar, XPG, and TeamGroup.

This week at Computex 2026, we saw Nvidia reveal its RTX Spark, and last month, AMD detailed its Ryzen AI Max 400 "Gorgon Halo" lineup, a refresh of the Strix Halo APUs that lifts supported unified memory to 192GB and allows up to 160GB of that pool to be addressed as VRAM. AMD describes the flagship Ryzen AI Max+ PRO 495 as the first x86 client processor able to run a 300-billion-parameter language model locally, pitching the platform for use cases that need to keep multiple AI agents resident in memory at once.

The market for Gorgon Halo will likely be directly shared with other chips, such as Nvidia's RTX Spark, which debuted at Computex 2026. RTX Spark is also positioned as an on-device agentic computing device. With local AI computing demanding lots of on-device RAM, it poses a difficult issue for device vendors.

DRAM contract prices are forecast to climb another 58% to 63% this quarter, on top of the record 90% to 95% jump TrendForce recorded in Q1, which also saw Nvidia raise the price of its DGX Spark desktop from $3,999 to $4,699, citing memory supply. So, what happens to the dream of accessible local AI compute?

DRAM supply squeeze

The local AI PC has become a category defined by how much memory it carries, and it’s scaling that memory up at a time when memory has never cost more. AMD's three Gorgon Halo SKUs reuse the same Zen 5 cores, RDNA 3.5 graphics, and XDNA 2 NPU as the existing Ryzen AI Max 300 parts, with the Max+ PRO 495 gaining a 100 MHz boost-clock bump to 5.2 GHz, a 40-compute-unit Radeon 8065S, and a 55 TOPS NPU.

Memory capacity has been increased 50% from the 128GB ceiling on Strix Halo, with a leaked PassMark entry putting the 192GB figure as eight 24GB SK hynix LPDDR5X packages on an HP test board, though AMD hasn’t yet confirmed this. Partner systems from Asus, HP, and Lenovo are due in the third quarter of 2026.

It’s all well and good that Nvidia and AMD are releasing machines like the RTX Spark and the Gorgon Halo line-up. However, Samsung, SK hynix, and Micron have all shifted the bulk of their wafer capacity toward high-bandwidth memory for AI accelerators because HBM carries far higher margins than commodity DRAM, and the conventional memory supply has tightened as a direct result of this. HP told investors in February that memory now accounts for roughly 35% of the cost of building a PC, up from 15% to 18% a quarter earlier.

SK Group chairman Chey Tae-won, speaking at Computex 2026 on the show’s official opening day, repeated his position that the shortage will run through 2030, despite the company's intention to double wafer capacity within the next five years. New fabs from all three makers are under construction, but none will reach volume production before late 2027 at the earliest, and most forecasts now predict a structurally higher price floor that persists even after the acute shortage eases.

The 192GB in a Gorgon Halo box, the 128GB in an RTX Spark or DGX Spark, and the LPDDR5X soldered into every AI laptop announced at Computex all come off wafers the memory makers would otherwise sell as HBM. That’s why Nvidia raised the DGX Spark by $700 in February without changing a single spec, and why component makers have begun passing memory costs through directly. One vendor has even taken an extremely on-the-nose approach of adding a flat memory surcharge to every purchase, and in some cases, smaller buyers are now quoted prices that change by the hour.

Bandwidth caps inference speed

A single pool of 192GB would enable an APU to hold a model that would otherwise require a multi-GPU server. While it doesn’t make the model run quickly, dense language model inference reads close to the full set of active weights from memory for every token generated, so generation speed is governed by memory bandwidth divided by the per-token weight footprint, not by idle memory.

Gorgon Halo keeps the same 256-bit LPDDR5X-8000 interface as Strix Halo, which tops out around 256 GB/s in theory and which independent testers have measured closer to 212 GB/s on the GPU. By comparison, the Apple M3 Ultra that AMD and Nvidia are chasing on capacity is rated at 819 GB/s, and an RTX 5090 moves data at 1,792 GB/s.

This gap explains why a dense 70-billion-parameter model fully resident on a Strix Halo iGPU lands in the low single digits of tokens per second, regardless of how much headroom the memory pool has. Our own Corsair AI Workstation 300 review found that Nvidia's slightly higher-bandwidth GB10 pulled ahead of Strix Halo as context length grew, for exactly this reason.

Capacity matters most for mixture-of-experts models, which activate only a fraction of their parameters per token and run far faster than their total size suggests, and for long-context agentic workloads, where it’s KVcache rather than model weights that consume memory. It’s these use cases that AMD’s agentic pitch points at, with leaked details on the next-gen Medusa Halo parts showing a move to LPDDR6 and as much as 80% more bandwidth.

Holding the line on price

Agentic AI is also something of a pricing tool for vendors, beyond describing a workload. A 192GB workstation sold on the promise of running 300-billion-parameter models locally can hold a four-figure price more comfortably than a mini PC sold on cores and clocks, and it justifies loading the most expensive component in the build to its maximum. AMD's Ryzen AI Halo developer box, a 128GB Strix Halo system, opens pre-orders in June at $3,999 through Micro Center, matching the launch price of Acer's GB10-based Veriton GN100 and the original DGX Spark before its increase.

Apple, the one vendor with the scale to hold priority memory allocation, has moved the other way. It pulled the 512GB Mac Studio configuration from sale, raised the price of its 256GB upgrade, and in May removed several more high-memory Mac mini and Mac Studio options as supply tightened.

This shows us beyond doubt that expanding capacity while holding the line on premium pricing is a choice the AMD and Nvidia camps are making, not one that the market is forcing. Whether buyers accept it rests on whether local agentic inference delivers enough value over cloud services to justify the outlay, on machines shipping with memory capacities that outpace the bandwidth that ultimately determines what that memory can do.

Astera Labs demonstrated its recently introduced Scorpio X-Series 320 Lane Smart Fabric Switch, which appears to be the industry’s largest open memory-semantic fabric switch, at Computex 2026 in Taipei. The PCIe 6.0 switch with 320 lanes can be used to build large multi-GPU scale-up clusters, large shared KV-cache memory pools, and disaggregate data center infrastructure using custom topologies.

The switch provides 320 PCIe 6.0 lanes and 20 Tbps of switching bandwidth, up from 144 lanes and 9 Tbps for previous-generation devices. Astera Labs says the increased number of lanes enables larger scale-up domains, enabling the connection of up to 80 accelerators using a single switch. By contrast, older 144-lane switches support up to 32 accelerators per switch. For clusters with more than 64 accelerators, the company says the new device reduces switch hops from as many as three to one and cuts switch count by a factor of four to six while still providing all-to-all connectivity akin to that provided by Nvidia’s NVL72 systems (albeit with lower bandwidth and higher latencies). The switch can support both standard and custom accelerators as long as they use standard PCIe connectivity.

On the trade show floor, Astera Labs is showing off its switching capabilities with Intel’s Arc B70 Pro graphics cards. However, real-world deployments based on the Scorpio X-Series 320-lane PCIe switches will likely use more advanced Intel hardware. In general, the switch can be used to build clusters from all types of accelerators that do not support their own NVLink or UALink-like interconnections, including AMD’s Instinct MI350P and Nvidia’s RTX 6000 Blackwell. Astera has yet to showcase a full working cluster featuring 80 accelerators, as the company only got the Scorpio X-Series 320-lane PCIe switch from the fab eight weeks ago. Also, finding 80 similar accelerators is not easy. Nonetheless, based on the company’s demonstration, the switch appears to be working.

A key feature of the Scorpio X-Series is Hypercast, which is a hardware-based data replication engine intended to accelerate communication-intensive operations common in AI models. According to Astera Labs, MoE networks tend to route tokens across hundreds of experts and create large amounts of multicast traffic between accelerators. In such cases, traditional switching architectures either require repeated data transmissions or slow multicast-group reconfiguration, whereas Hypercast is designed to handle these communication patterns directly in hardware, reduce GPU networking overhead, and improve accelerator efficiency, Astera Labs claims.

The company also added In-Network Compute engines that offload collective operations such as AllReduce, ReduceScatter, AllGather, AllScatter, and all-to-all exchanges. These features can reduce communication latency by more than 50% in certain workloads, according to Astera.

Another important feature of the Scorpio X-Series 320-lane PCIe switch is its memory-semantic connectivity, which enables connected processors to access fabric-attached resources using native load and store operations rather than software-controlled transactions. This greatly simplifies usage of the device and improves real-world performance by reducing overhead and improving fabric efficiency at scale.

Astera says that the production ramp of the Scorpio X-Series 320-lane PCIe switch is set for the second half of 2026. Currently, the company is sampling the switch with leading hyperscalers.

Cherry's gaming branch, Cherry XTRFY, introduced the first 8K ultra-wideband gaming keyboard at Computex 2026 earlier today. The new K63W Pro features a 70-percent layout, gasket mount construction, and low-profile switches for a slim, compact, gaming powerhouse with ultra-low latency and high throughput. The keyboard will debut in the EU in early July for €179.99 and will launch in the U.S. in August at $169.99.

It's not the first 8K gaming keyboard, but it is the first with ultra-wideband technology — a short-range, high-frequency wireless technology that's less prone to signal interference than 2.4GHz wireless. This means the keyboard will have a more stable high-speed connection that won't mess with (or be messed with by) other wireless peripherals, such as wireless gaming mice or headsets. The keyboard gets true 8,000 Hz polling over both its wired and wireless connections, which is still probably overkill for most users but means it should (in theory) be the fastest, most responsive gaming keyboard on the market when it launches.

“With the K63W Pro, we wanted to evolve wireless gaming even further. Ultra-Wideband technology and 8K polling deliver extreme performance, while the carefully tuned gasket construction creates a typing feel you would never expect from a low-profile keyboard. The result is a keyboard built for both competitive gaming and hours of daily typing,” Joakim Jansson, Director of Product Management at Cherry, said in a press release.

Design-wise, the K63W Pro is a 70-percent keyboard — meaning it keeps the full function row and arrow keys, but drops the single-key navigation row you see on the more typical 75-percent layout. Cherry says this new layout is "optimized for gaming, keeping the F-row while reducing side bulk to maximize space for mouse movement." In addition to a modified layout, the keyboard is also low-profile — featuring Cherry's MX Low Profile 2.0 switches in a gasket mount design, which creates a typing experience that "feels controlled, cushioned, and surprisingly deep for its size" (according to Cherry).

The keyboard also has a 6,000 mAh battery that delivers up to 1,100 hours of usage, "depending on settings such as polling rate and RGB intensity." We'll assume this means that it won't get 1,100 hours over an 8,000 Hz polling rate, but who knows — Keychron's ZMK-based Q Ultra series gets up to 660 hours of battery life over 8,000 Hz polling rates, so maybe the K63W Pro will surprise us.

The Cherry XTRFY K63W Pro will launch in the EU in July and will hit shelves in the U.S. in August, with retail prices of €179.99 and $169.99, respectively.

Intel knows that Arrow Lake dealt a huge blow to its reputation among gamers and enthusiasts. The underperforming chips sporting a radically different architecture didn’t make the cut in games, underperforming in some titles compared to even last-gen parts, and although application performance is competitive, it isn’t enough to earn a spot among the best CPUs. Tom’s Hardware asked Intel’s Nish Neelalojanan, senior director of product management for Intel’s Client Computing Group, about the low prices of the newer Core Ultra 7 270K Plus and 250K Plus at Computex 2026, and he told us the prices were chosen to help build back Intel’s reputation.

We’ve presented this same question to Intel multiple times before, and always with the same response. Fundamentally, the Core Ultra 7 270K Plus undermines Intel’s existing Arrow Lake range. It’s faster in games and applications compared to the Core Ultra 9 285K, and it costs half the price. So, we pressed Neelalojanan further about this issue in an attempt to understand why exactly Intel decided to target such aggressive pricing.

“From an enthusiast perspective, it was... we needed to build back our reputation. I am sure you would agree with that, and this was, hey, [we’re] making sure we are providing value to the gamers, and we start with Arrow Lake Refresh, and we have a very strong roadmap to come,” Neelalojanan told me. “In terms of desktop, that was an effort to [say] ‘let's go with value focus first, and that will help us then gain confidence.’ I mean, this is obvious... I'm not saying anything which you already don't know.”

The next step in that roadmap is Nova Lake, which is a generation Intel has been talking about for quite some time. The company has reiterated that Nova Lake is set to launch at the end of 2026, but it’s not clear if we’ll have actual products by then. Speaking to a different Intel representative, they told me that demand for CPUs spans everywhere from the data center down to mainstream laptops, and that the company sometimes has meetings on a daily basis to address these changes. The rapidly evolving market around CPUs could push Nova Lake out to next year, but, at the moment, Nova Lake is still slated for the end of the year.

It seems like each new generation that isn’t a refresh from Intel is some make-or-break moment, but outside of the hype, Nova Lake has a lot of interesting murmurs surrounding it. First and most obvious is bLLC. We’ve heard continual rumors that Intel is working on a “big last-level cache” in a bid to finally combat AMD’s X3D chips, and apparently, that will show up with Nova Lake. Intel hasn’t confirmed that’s the case yet, of course.

Nova Lake has a different context compared to the launch of Arrow Lake Refresh, as well. My conversations with Intel point to a reset in the company’s enthusiast roadmap, as it clearly knows the damage Arrow Lake did to its reputation. We’re still far off from Nova Lake pricing discussions, but given the aggressive rollout of Arrow Lake Refresh, I wouldn’t be surprised to see Intel undercut AMD across the spectrum.

Neelalojanan certainly implies that’s the case. We’ll need to wait until Intel has more to share on Nova Lake, which hopefully we will hear about soon. Again, the company has said Nova Lake is on track for the end of 2026, so barring a delay announcement, that’s when we expect to see the chips in the flesh.

Computex 2026 is moving full steam ahead, like an AI train running down a track made of gold-plated DDR5 DIMMs. We’ve moved into the first full day of the trade show, and the announcements are really starting to fill in. You can catch our Day Zero coverage here and keep track of our dedicated Computex 2026 hub.

Arm PC chips are back in focus

Windows on Arm is not new; the first-generation Surface RT launched way back in 2012 with an Nvidia Tegra 3 processor. Since then, we’ve seen various other takes on Arm processors running on Windows, from the Surface Pro X with its Microsoft SQ1 SoC to a slew of laptops running Qualcomm’s PC-centric Snapdragon processors.

Now, we’re seeing an even more interest in the segment with Nvidia RTX Spark “Superchip” and the new Snapdragon C from Qualcomm. From all accounts, the RTX Spark targets the high end of the PC market with its 20-core Arm CPU, Blackwell RTX GPU, and 128GB of unified memory.

The Snapdragon C takes a different approach, instead aiming at the budget laptop segment. Laptops using Snapdragon C are expected to be priced as low as $300. However, that price point will be highly dependent on memory pricing, which remains a real pain point not only for OEMs, but also consumers looking to get the most bang for their computing buck. In fact, things have gotten so bad that even Intel has said that "something has to give” with memory prices.

• Nvidia's RTX Spark could capitalize where Qualcomm's Arm-based efforts have not
• Jensen Huang says Nvidia wants to 'reinvent the single most important tool of humanity' with RTX Spark
• Qualcomm aims the Snapdragon C laptop chip at the budget laptop segment, as manufacturers feel the DRAM squeeze
• Intel says 'something has to give' with memory prices

We get hands-on time with Asus’ latest hardware at Computex

Asus always has a large presence at Computex, and this year was no exception. The company had an extensive cast of new characters in the laptop field, with new Vivobooks, Zenbooks, Expertbooks, and Strix Scar gaming laptops. Of the new models introduced, the Zenbook 14 with an Intel processor and 14-inch 2880 x 1800 OLED display caught my attention.

Asus also had a rather cool-looking ROG 20th anniversary Harpe II Extreme Edition wireless mouse and the ROG Azoth Extreme Edition 20 mechanical keyboard. The devices are finished primarily in black, but feature 24-karat gold accents. The keyboard also includes transparent switches and keycaps. And we can’t get past its 3.5-pound heft, thanks to its all-metal chassis.

• Asus shows off updated Zenbook and Strix Scar laptops, along with a TUF-based Gaming desktop
• Hands-on with Asus’ ROG Harpe II Extreme Edition 20 gaming mouse
• Hands-on with Asus’ ROG Azoth Extreme Edition 20 mechanical keyboard

PCIe 6.0 SSDs are on the horizon

A year ago at Computex 2025, we saw prototype PCIe 6.0 SSDs capable of 30+ GB/s speeds. This year, we’re slowly inching towards production-capable hardware, with Phison showing off its latest PCIe 6.0 SSD controller: the X3. The X3 is a 16-channel design that is capable of 28 GB/s sequential read/write speeds and 6.8 million random read/write IOPS.

According to Phison, the controller will begin sampling to customers by the end of 2026, with volume production starting in mid-2027.

• Phison shows PCIe 6.0 X3 SSD controller with 28 GB/s of bandwidth and 6.8 million IOPS, supports 2 petabytes per drive

Everything else

We can’t get into detail on everything we’ve seen so far at Computex in this wrap-up — we’ll let the individual news stories speak for themselves. Here’s everything else we’ve covered for Computex 2026 Day One:

• Noctua announces new thermal pad for AMD chips in partnership with Carbice
• Gigabyte debuts fourth-gen Tandem WOLED and multi-mode Mini LED gaming monitors
• Gigabyte showcases new Infinity products for its 40th anniversary
• Cooler Master shows off new HAF 500 chassis, aluminum fans, and new air coolers
• AMD ‘had to re-engineer’ the Ryzen 7 5800X3D for a re-release

Mainland Chinese companies among the 219 listed mainland exhibitors at Computex 2026 in Taipei have been kept off the show floor by stalled entry permits, the South China Morning Post reported today, as the four-day trade show officially opened. Staff at listed exhibitors reportedly told the publication that nobody on their teams obtained an entry permit from Taiwanese authorities this year, with applications either left pending or met with last-minute requests for documentation that was difficult to supply on short notice. No formal rejections have been issued, mirroring a similar permit freeze that hit April's Taipei AMPA auto-parts show.

SCMP spoke to two mainland residents at listed exhibitors and a third mainland citizen employed by a multinational, all speaking anonymously, given the political sensitivity, who described their applications stuck in limbo while colleagues from other regions were given access. Two travel agencies that handle cross-strait trips told SCMP that none of their clients, including official exhibitors, secured approvals this year.

Emdoor, a mainland VR and electronics maker that has exhibited at Computex for more than a decade, was an exception and said it obtained permits without difficulty. Taiwan's National Immigration Agency, which processes the applications, told SCMP it "reviews such applications in accordance with established procedures and consults with the relevant authorities."

Mainland nationals need travel documents from both Beijing and Taipei, and the people who spoke to SCMP said the Beijing side wasn’t the holdup. Since 2023, mainlanders have been able to apply directly to the National Immigration Agency for short-term exhibition permits, but anyone whose sponsor or activity touches high-tech sectors such as semiconductors must first secure separate special approval, according to advisories from immigration law firms Fragomen and BAL.

Those advisories also note that the number of staff a company may send is capped under revenue-based quotas. That being a discretionary decision, rather than an outright ban, is what allows Taiwanese authorities to indefinitely delay an application without having to officially reject it.

Taiwan has been taking an increasingly harder stance on mainland access to its tech sector this year, having launched its first formal crackdown on illicit AI hardware exports to China in May. Taiwanese courts are also currently prosecuting TSMC trade-secret theft under the National Security Act.

We’ve also seen several reports this year describing intensifying Chinese efforts to poach Taiwanese chip talent, set against more than 420 Chinese military aircraft operating near the island in the first quarter. Mainland travel to Taiwan has fallen accordingly: 58,708 mainland residents visited in the first quarter of 2026, down 13% year over year, according to Taiwan's interior ministry.

AMD just launched the Ryzen 7 7700X3D, which is a strange entry in AMD’s AM5 lineup given that it’s built on the Zen 4 architecture rather than Zen 5. We asked AMD’s David McAfee, VP and general manager of Ryzen and Radeon, at Computex 2026, why the company went with Zen 4 over a six-core Zen 5 part. After all, we’ve seen the six-core Ryzen 5 7600X3D but not a Zen 5 replacement. McAfee told us that “may be something that we look at doing… later this year.”

“I think as we go through the rest of this year, I mean we're always looking for ways to, you know, create as many options as we can, and that may become… maybe something that we look at doing as a runway product later this year, simply because you know we know the pressures that are there in building systems aren't going away anytime soon,” McAfee told Tom’s Hardware.

A lot of hedging is required here because McAfee didn’t confirm that the Ryzen 5 9600X3D is in development, nor that it will, for sure, come this year. But it’s clearly something AMD has thought about, and the decision not to release a six-core part at this time came down to more than die allocation.

“The 7600X3D, that has been more of a limited run product, where it really isn’t, you know, day-in day-out available from all of our partners,” McAfee said. “From a supply standpoint, I’m not going to say it’s more difficult to make, but one that doesn’t naturally occur as much as an eight-core product. So, we have been a little bit more selective with how we’ve used the six-core products.”

Broadly, however, McAfee says the decision to release the Ryzen 7 7700X3D came down to a preference from gamers for an eight-core chip, despite recognizing that “the difference between eight cores and six cores isn't huge in terms of gaming,” at least across a broad range of titles. Certain games will see a performance loss with six cores as opposed to eight, but they’re few and far between.

We saw that in action in our recent review of the Ryzen 5 7600X3D, in fact. Despite being nearly two years old, the chip holds up surprisingly well in today’s market at its $230 price on Amazon ($200 in-store at Micro Center). In our testing, the Ryzen 5 7600X3D is just 4.5% slower than the Ryzen 7 7800X3D on average, despite sporting a massive cut to TDP, two fewer cores, and less cache.

Given the context of six-core X3D parts more broadly, if AMD decides to introduce the Ryzen 5 9600X3D, it likely won’t be a far-reaching chip like the eight-core parts are. The Ryzen 5 7600X3D remains a Micro Center exclusive chip, though the retailer recently started selling directly on Amazon, opening up availability for enthusiasts that don’t have a physical retail location nearby.

That has vastly lowered the barrier to entry with X3D CPUs, especially as Zen 3 options have dried up in the market. Prior to the Ryzen 5 7600X3D showing up online, you’d have to spend at least $350 (and likely more) to get an X3D CPU. The Ryzen 7 7700X3D pushes that entry point down further at its $330 suggested retail price. However, an eight-core part, assuming a similar price range between $200 and $250, would push that barrier to entry even lower.

McAfee recognizes how that could benefit gamers in this supply-constrained environment of high memory and storage prices, pointing to the benefits of a large L3 cache when running slower, or even single-channel, memory.

Cooler Master had a lot of things to show off at Computex 2026. The keynote, held at their headquarters just up the road from the Nangang Exhibition Center, talked about their AI data center cooling and eventually delved into the consumer side of things where we live. After the keynote, we headed downstairs to a huge room showcasing not only the consumer side of things but also the CDUs (Cooling Distribution Units) they’ve developed. It was quite a display if you’ve never seen them before.

But alas, we’re here to talk about a couple of new cases in the HAF II 500, and Silencio 600, the new V8 air cooler, and the all-aluminum Masterfan A series fans debuted here.Starting with the HAF II 500, this High Airflow mid-tower chassis is purpose-built for high-end systems, including gaming, creator, overclocking, and even AI systems that may require better cooling. Two huge 220mm intake fans at the front, purpose-built for the case (may sell separately), and a large 180mm exhaust fan (all from the Mighty 40 family), the case helps live up to the HAF name. While we don’t have any cooling metrics, I can tell you that you could feel the airflow from behind the case from a few feet away, and we couldn’t hear it (admittedly, over the hum of the crowd). But don’t worry, we’ll get in the lab as soon as it's available and put it through its paces.

HAF II 500

Internally, it supports up to E-ATX motherboards, so you can put just about anything inside and keep it cool without making much noise. The ribbed front panel’s low airflow resistance helps air enter the chassis and exit the rear with minimal restriction. A portion of the bottom fan cools the PSU chamber, while most of the airflow is directed by a curved baffle to cool the GPU, where it’s needed most. On top of that, it uses an evolution of the Freeform 2.0 platform and sliding rails, called the MasterRail System, a convenient way to support multiple fan and radiator sizes without blocking airflow.

The HAF II 500 supports dual-GPU setups up to 3.6 slots thick across the eight expansion slots. In other words, there’s plenty of room to put whatever you need inside. The outside is your typical chassis with a tinted glass side panel to show off the internals. It uses split-level cable management with zoned routing space, integrated straps, and a cable cover that swings out for ease of installation or part swapping. The front panel’s ribbed design allows for ample airflow. We should see the HAF II 500 around July, priced between $179.99 and $209.99.

Keep an eye out for a review in the coming weeks. The next case we saw was the updated Silencio 600. As the name implies, this mid-tower case is made for silence. Take a look at the unique front panel, and you can instantly see that it’s different from most other cases and from the previous Silencio 600. This unique design closes the gaps with a soft-form sound-deadening material to keep internal system noise in while maintaining front-panel airflow. Behind the panel are two pre-installed 180x40mm fans to keep the air moving and your internals cool, even at lower RPMs, with less noise. The silent case is Fit for shared spaces, home offices, studios, or anywhere else that needs a quiet case.

Silencio 600

We don’t have many other details on the chassis, including internal specs, but it will be available around October of this year (no MSRP, either).

Metal fans seem to be a theme at Computex 2026, and Cooler Master wasn’t missing out. The new MasterFan A series uses an aluminum frame and blades to increase performance. Due to increased material rigidity, the blades have less flexibility, enabling higher speeds and CFM. The blade-to-top gap on the A120 is 0.6mm and 0.8mm on the 140mm. While that doesn’t trump Noctua’s tolerances, it’s a lot closer than most and still yields improvements both acoustically

Masterfan A

The A120 boasts over 80 CFM and 6.1 mmH2O, while the A140 moves 104 CFM at almost 3.5 mmH2O, which is a fair amount of static pressure, especially for the A120. The fan comes in two flavors: the base model, which runs up to 2,500 RPM.. It uses a three-phase motor with dual ball bearings. Just keep your fingers away from the blades! These should be available soon, rolling out globally, with pricing around $24 for the A120.

Last but certainly not least is the V8 Ace 3DHP. Keeping the same bold, automotive engine aesthetic from the original V-series, the new flagship cooler gets updated 3DHP (3D Heatpipe) technology from the data center side of the house, said to improve heat transfer and maintain performance during intensive workloads. Cooler Master says its single-stack, dual-fan configuration is still competitive with much larger, dual-stack towers like the Noctua NHD015 G2, but at a lower price point (not listed yet), making it an attractive offering when available (soon!).

V8 Ace 3DHP

There are two versions, one for AMD and one for Intel, and they work a bit differently. The AMD version has less internal water volume in the heatpipe than the Intel SKU, which uses more water to stave off temperature spikes from the higher-wattage chip below. Both use the ‘trident’ heatpipe configuration, with a heatpipe rising into the middle of the fin array, reducing the thermal dead zone found in typical “U”-shaped heatpipe designs and increasing heat extraction by 30%, according to Cooler Master.

All in all, we saw a lot of ‘cool’ things at Cooler Master HQ at Computex 2026, and I personally can’t wait to test them.

Phison's booth at Computex 2026 had its new PCIe 6.0 SSD controller, dubbed the X3, on display, with claims of up to 28 GB/s of sequential throughput and 6.8 million IOPS in random read/write workloads. The company also had performance benchmarks for its new DRAM-less PCIe 5.0 SSD controller, the E37T, on display, demonstrating similar performance to its popular flagship E28 controller while sipping nearly half the power, setting the stage for a new wave of power-efficient SSDs that will run cool.

Phison's 16-channel X3 controller is now nearing completion. The firm demoed this same PCIe 6.0 x4 controller at CES on a large test validation motherboard, but now it is finalized enough to be crystallized down into reference SSD designs, signaling it is almost ready for sampling. The company tells us the controller will sample to customers in December, and then ship in volume in mid-2027.

The controller is spec'd to deliver 28 GB/s of sequential read/write throughput and up to a blistering 6.8 million random read/write IOPS. Peak storage capacity weighs in at an incredible two petabytes; yes, 2 petabytes per SSD. Power efficiency is a focus for this controller, with Phison claiming 4 GB/s per watt, which works out to a total power draw of 7 Watts. The controller supports all the latest specs, like NVMe 2.3, OCP v2.6, and a full suite of security features. Phison's two reference designs come in data center-focused E3.S and E1.S form factors, but it's logical to assume that we'll see variants of this in the consumer M.2 form factor in the future.

As you can see in the album above, Phison also has a full suite of PCIe 6.0 redrivers, retimers, and cabling being readied for launch, giving it a robust suite of PCIe 6.0 IP for the next wave of storage devices and systems.

Phison also had its reference SSD design with a DRAM-less PS5037-E37T controller up and running on a laptop. The demo sported 14,239 MB/s of sequential read throughput and 12,307 MB/s of sequential write performance, which is roughly equivalent to the speed of its flagship DRAM-equipped E28 controller. The drive is also spec'd to 3 million random read/write IOPS, all of which comes courtesy of 4800 MT/s BiCS NAND. That top-tier NAND isn't available on the market yet, but it is clear that it is headed to shelves soon.

The company says this controller delivers 14.9 GB/s while sipping a mere 4.5W, a full 2.5W less than its flagship DRAM-equipped E28. Naturally, the DRAM-less design will also have cost savings attached, given that the price of any kind of DRAM is currently apocalyptic. The E37T will begin shipping this year.

Intel launched its Xeon 6+ "Clearwater Forest" processors at Computex 2026 in Taipei, and on Monday, two of the individuals responsible for the product sat down with the press to answer questions. Kira Boyko, Product Line Director for E-Core Xeon Products in Intel's Data Center Group, led the session, which was joined partway through by Tim Wilson, Vice President and General Manager of Intel's Data Center Silicon Engineering group.

Across roughly half an hour, the two addressed why Intel stripped hyper-threading out of its E-core server parts and the technical case for bringing it back, the agentic AI demand surge that has left expensive GPU fleets idling while they wait on CPUs, the deliberate decision to ship Clearwater Forest with only AVX2, and 18A supply so tight that allocating chips between customers is "daily, in some cases."

Diamond Rapids, Intel's next P-core Xeon, drew repeated questions, but Intel deferred any detail, with Evangelista pointing reporters to fuller commentary roughly two months out. That timing lines up with Hot Chips, where Intel is expected to share more on Diamond Rapids.

Clearwater Forest spec changes

Clearwater Forest tops out at 288 Darkmont E-cores per socket and 576MB of L3 cache, and is Intel's first data center CPU built on its 18A process.

Kira Boyko: It's our most performant Xeon on the market today, specifically for scale-out workloads, so it's not just a per-watt angle of fossil performance.

Jake Roach, Tom's Hardware: Was that the driving force behind the big spec changes compared to Sierra Forest? Obviously, it's double the core count, but I think there's over five times the amount of L3, and a huge increase in TDP.

Kira Boyko: The TDP is mostly that it is socket-compatible with the version of platform design that we had for Granite [Rapids]-AP before, and that is a higher-TDP product. Our initial E-core part was lower TDP, and this one has roughly the same range as the Granite version, so that's part of the platform-design alignment. But in general, we found that our customers were mostly targeting higher-TDP spaces anyway for the core density they were after, so it ended up working quite well. We already had a design that served those spaces.

Jake Roach: And the L3, was that another workload type?

Kira Boyko: You're right, a little over 5x increase, from the hundreds up to 576-ish [MB].

Jake Roach: If you have the flagship, that's quite a lot of L3.

Diamond Rapids and hyper-threading

On its recent earnings calls, Intel CEO Lip-Bu Tan has said that moving away from simultaneous multi-threading (SMT) "put us at a competitive disadvantage" and that the company will reintroduce it with the Coral Rapids generation, the P-core Xeon that follows Diamond Rapids. Intel's current shipping E-core Xeons, Sierra Forest and Clearwater Forest, run a single thread per core.

Jake Roach: I appreciate that we can't comment on future products, but there was a tease in the press deck for Diamond Rapids. It's a very anticipated product. I want to ask about hyper-threading. During the last two earnings calls, Lip-Bu Tan has referenced that hyper-threading will return with Coral Rapids. We don't currently have a Xeon P-core product without hyper-threading shipping. Does that mean Diamond Rapids does not have hyper-threading?

Andrew Evangelista: We'll comment more on Diamond Rapids [later].

Kira Boyko: I will say that E-core is single-threaded. It has the core density for the workloads it's servicing, and we are not expecting it to be replaced by Diamond Rapids. We're expecting the workloads that need more of the high-performance aspect of a P-core to go from Granite to Diamond, whereas the more scalar workloads for E-core will stick on Clearwater Forest, continue to be serviced through this next generation, and then pick up with the generation after that.

Andrew Evangelista: I think it was your question that prompted us to talk about Diamond Rapids in general. More to come.

Kira Boyko: The only thing I can really say from a Diamond perspective is AET, the new feature we're introducing on Clearwater Forest. That is expected to roll out across all of our Xeons going forward, so you can expect to see it on Diamond and future ones as well. What the roadmap looks like from a feature perspective gen to gen, I don't have any level of detail for today. But it is definitely being introduced in Clearwater. We have a number of customers deploying it on Clearwater, and others who are more classic P-core customers, with different workloads, running proof-of-concepts on Clearwater so they can hit the ground running with Diamond.

Jake Roach: I had to ask.

Kira Boyko: You're like, "That's not what I wanted, I wanted way more detail."

Jake Roach: There are a lot of articles that have been written saying it's confirmed Diamond does not have hyper-threading, and I haven't been able to find that confirmation anywhere.

Kira Boyko: There's a lot out there on Diamond that is all over the place. There are statements about [unclear] variants, and just a lot of rumors.

Andrew Evangelista: We'll have more official commentary to come, and that's what folks are anticipating. We understand there's excitement, and we'll share more [in two months].

Agentic AI and CPU demand

Journalist 2: What are customers telling you about this agentic AI wave, how they're dealing with it, and how Intel plays into that going forward?

Kira Boyko: Customers are just starting to understand their own AI deployment models, and a lot of them still aren't quite there yet. Many started by investing in GPUs and are now realizing they don't have the CPU counterparts to actually keep those GPUs going. So they made this huge investment, and they're running at something like 20% to 30%, something quite low. They're understanding that there's this space where certain workloads can be offloaded to more efficient CPUs, and that's exciting from a Xeon 6+ standpoint. Others are still going to be partnering with their providers, looking at industry white papers to understand how to best use their AI strategy.

Journalist 2: Just as an outside observer, it seemed like CPU demand was going, and then November and December happened, and everything got sold out instantly. What percent of current demand is agentic-AI-driven versus prior? I'm trying to get a sense of what it is now, and what it's going to be like six to nine months from now. It seems like a paradigm shift happened, and we're going to be riding this trend for several quarters at least.

Kira Boyko: I think we are. I think we're also going to see quite a bit of data center modernization and consolidation, looking at what workloads are already out there that can be consolidated onto CPUs. Some maybe are designed for agentic, maybe aren't, but are more storage-oriented, or workloads that can be serviced just fine on something that isn't super intense. So you can get a little more performance and energy back, and then use that to service some of their AI workloads as well.

Application Energy Telemetry

AET, or Application Energy Telemetry, is a Clearwater Forest feature that gives operators application-level visibility into energy use, which Intel says can be used both to tune workloads and to bill customers on measured rather than estimated consumption.

Jake Roach: You mentioned AET, and I know that's a really big thing with this launch. Is there any connective tissue with what we saw with Arrow Lake Refresh on the consumer front? AET is taking information from actual registers in the silicon. There's hardware on the chips doing it. Similarly, with iBot on Arrow Lake Refresh, it was hardware-enabled, where you could get these readouts running workloads and see where they could optimize. Is there any connective tissue there, or are these completely separate?

Kira Boyko: We can get back to you on whether there's some collaboration. Usually our teams are very separate, but it's very possible there is some, so we'll find out. My understanding is that this is highly customer-driven. Sometimes we leverage existing technologies.

Jake Roach: This is more to satiate my own curiosity.

Andrew Evangelista: Let me grab Tim for a second to answer that, because that's a silicon-engineering-level question. He's worked on both client and enterprise.

Jake Roach: Yeah, just because you're using Darkmont, there's at least a capability there.

Kira Boyko: Touché. Moving on to the next generation, which is not an E-core, it'll still be there as well. So even if there was some synergy, it would be moving forward to a different core base.

Andrew Evangelista: Circling back on two questions. One was similarities related to Arrow Lake.

Jake Roach: Basically, the hooks in Arrow Lake Refresh for iBot to optimize that translation. Are those hardware hooks something you're looking at? Is there any connective tissue there today?

Tim Wilson: I haven't looked at Arrow Lake in quite a while. To first order, I'd say fundamentally no, they're different use cases. Are we leveraging some of the same telemetry capability built into the hardware? It wouldn't surprise me.

SMT removal and its return

Intel split its Xeon 6 line into P-core parts with hyper-threading, such as Granite Rapids, and E-core parts without it, beginning with Sierra Forest.

Andrew Evangelista: The question on the decision for SMT and hyper-threading.

Journalist 2: Just why it was taken out, what the thinking behind that was. Was it a security thing?

Tim Wilson: I have a lot of my own personal thoughts. If you step back to the data center a couple of years ago, the thesis was that what matters is maximum core performance and then core density in the socket. So if I can deliver maximum core performance and increase the number of cores in the socket, do I really need [SMT]? One or two physical cores are always better than two virtual cores built on one physical core. I fully expect we will see use cases and workloads where that decision is incredibly useful and valuable and gives real-world value, and we've heard from some customers that, for what they're doing, single-threaded is the right answer. Having said that, there's a big portion of work that is still very much multi-threaded, especially in the virtualized space, so completely eliminating it is a problem, because we cut off some not-insignificant portion of workloads, especially when you're in a virtualized, licensed environment where licensing is based on cores and threads. So there was a technical reason for why you'd want [SMT], and that technical reason probably still holds. [...]

Journalist 2: Like a VMware thing, with a big price on cores. How are you going to market versus your main rivals, like AMD, and now Nvidia? What's the messaging going forward? From 18A to what's next, you'll be much better on the node side.

Tim Wilson: Our intention is to have leadership products with every generation, and we fully intend to do that going forward. Our go-to-market strategy is to sit down with customers and ask what they value, then go build leadership that meets those needs. With all our data center customers, we have deep discussions around the personality of the platform they want to build, beyond just cores and feeds and speeds. What is the system balance, the memory-to-compute ratio? How are they viewing multi-socket versus single-socket? What's the right number of cores for their workloads, for both private and public workloads, enterprise versus cloud? It's really sitting down in each of those and asking what the markets and customers buying our parts value, and how we optimize our products to meet their needs.

Journalist 2: Is that changing now with this agentic AI demand explosion? Are we going to see CPU racks with agentic AI as the primary use?

Tim Wilson: I'm sure you will, just like we've always built CPU racks. There are principles around CPU design that have always been true and will continue to be true. You want the highest-performance core you can build. Power efficiency is always going to matter as long as we're constrained by the amount of power you can bring inside a building's walls and the heat you can extract from them. Your memory-to-CPU harmonics, how much memory each workload takes, how much you allocate to each core, those are key. We've always designed for those parameters, and the end markets evolve over time.

Agentic AI is now exploding, but what's driving that explosion is not a new type of CPU. It's that the new AI workloads are not one call, one inference, one response. They're complex, execution-driven, multi-task queries that involve tens or hundreds of agents, and suddenly you need a control plane and an orchestrator, tasks the CPU is historically good at. How do I take a complex task and decompose it into subcomponents, figure out which can be parallelized and which depend on each other and need to be serialized, and pass those off to the GPU? I have to map memory to each of those subcomponents, and not all of them want the same memory, and I have to make calls to I/O, and in some cases to the OS or APIs.

Those are all things the control plane and orchestrator, the CPU, does really well. As you move away from a chatbot answer to "go do this analysis and give me a report on the actions I should take," that's a much different query, and the CPU plays a much bigger role. Data centers that have built on GPUs for the last three years are suddenly finding they're bottlenecked by the CPU. They have a massive GPU fleet that costs billions of dollars sitting idle, waiting for the CPU to respond. So do I see a future with agentic AI and CPU racks? Yes, but with characteristics very similar to the sorts of things we've always built into CPU racks. It's exploding because the things the CPU has always done well are the things in demand now.

Journalist 2: It seems like the whole storage infrastructure has to change, too.

Tim Wilson: That comes along with it. There's demand for storage, which drives I/O advancements and connectivity.

18A yield and wafer allocation

Clearwater Forest is a multi-process design: the compute tiles are built on Intel's 18A node, with base tiles on Intel 3 and I/O tiles on Intel 7.

Journalist 2: Questions you probably want to ask but won't answer. 18A yield volume for Xeon 6+, progressing?

Kira Boyko: We're ramping well. We have strong demand throughout the lifetime of the product, and we're working from a capacity standpoint across all of our products to hit customers at the point in time they need most. Compute is 18A, but we also have base on Intel 3 and I/O on Intel 7, so it's a multi-process product. We're mapping demand against all of our other products to figure out where we need to build.

Journalist 2: How do you choose who gets the product in this compute-supply-constrained world? Is it whoever orders first? You're sold out right now.

Tim Wilson: We give as many CPUs to as many people as we can. It tends to be more business decisions than engineering decisions, so it's a combination of long-term deals and customer relationships. The biggest problem is not demand in any way, shape, or form. The biggest problem is how we satisfy demand across every single product. If I have people demanding Xeon 6 and Xeon 6+, and still Xeon 5, how do I balance all of those and match where the supply constraints are in the industry? In some cases, customers are struggling with mismatches. They can get the GPU but not the memory to pair with it, or the memory but not the CPU. There's a lot of matching going on in the industry.

Journalist 2: On the client side, people are demanding even older products because they've already verified them.

Kira Boyko: I've seen that on the data center side, too. They've verified and tested a product, so they want that product. In such a supply-constrained environment, people will buy whatever’s on the table. And 6+ has the benefit of some backward-compatibility elements, the socket compatibility, and again, using processes that are hardened on previous products, so we can mix and match in some cases. We have customers looking for multiple products on multiple processes, and it's working with them to understand exactly what they critically need, when, and how we best service that across all their orders.

Jake Roach: If I'm remembering correctly, it was the earnings call before the most recent one, where we talked about wafer allocation split between client and data center, with a greater emphasis on wafers going toward the data center. Is that still the plan?

Tim Wilson: That's definitely the plan, and we're always having those conversations. That's more of a foundry conversation than a product conversation. [...] The whole ecosystem is sucking up all the wafers and memory, whether it's client, automotive, or any of the other industries. AI data center tends to take the supply because they're willing to pay the most, and the rest of the industries can't pay the price until supply balances out. We saw a similar effect during COVID, though that was supply-chain-driven rather than demand-driven. Those trade-offs, Gen 5 versus Gen 6 versus Gen 7, are a weekly conversation.

Kira Boyko: Daily, in some cases, on CPU allocations. [...] Given the dynamic space, our customers are modifying on a regular basis. Can we shift? What do they really need, and when? If you're asking long-term whether we'll stay in these constraints, we do see a space where things will lighten up. It's not in the immediate timeframe.

Journalist 2: Dave talked about multi-year hyperscaler contracts. What's the latest on that? Are deals getting signed, and are you getting more requests for those kinds of contracts?

Tim Wilson: I doubt either of us is the right person, by the way. They don't trust us with a lot of that information.

Journalist 2: You're not talking with data center customers on the purchase side?

Kira Boyko: We're not in the contract negotiation.

Tim Wilson: There's a principle here. If you mix commercial negotiations in with the technical discussions, it doesn't work out well, so you generally try to separate them. You let the finance people argue over pricing and contracts, and let the engineers figure out the best products to build together. When you're talking with product-side engineers, we don't have a lot of that information, and even if we did, we probably couldn't tell you.

Journalist 2: What percentage of Intel's data center revenue is hyperscaler?

Tim Wilson: You can go look at our earnings.

[Session ends]

Gigabyte’s press event at Computex 2026 showed off a slew of new items, including a monster 1,600W power supply, new monitors, AIOs, keyboards, motherboards, graphics cards, and a whole lot more we won't list in this article.

On the motherboard side of things, one of the first things it showed off is the X870 Aorus Infinity, sporting a unique two-DIMM design with the socket rotated 90 degrees and the two DRAM slots running above it. This design allows for ultra-high DDR5 performance, reaching a blazing 11,400 MT/s with a low CL24 rating. The E-ATX board has several buttons for overclocking and looks like a virtual twin to their overclocking board, the X870 Tachyon Duo X Ice. Pricing and availability were not mentioned (for any item) at the event.

The real star of the motherboards is the X870 Infinity Next. This halo board has a unique appearance thanks to the advanced 3D metal printing technology used to design the heatsinks, which resemble the pores of lava rock (or even a sponge) more than traditional metal blocks. The rocket thruster-grade thermal materials boast up to 44% greater cooling surface area than traditional designs.

It’s not all about the looks and cooling, though. This board is built like no other, offering 64 power phases using newly designed MOSFETs with Quad OptiMOS technology MOSFETs supporting up to 5,120A total current (so 80A each), almost double that of the next highest motherboard. Additional details are scarce, but this has to be one of the freshest, coolest-looking designs I’ve ever seen on a motherboard. This board is not cheap to produce, over $3,000, so it will only be in the hands of a few fortunate souls when it moves to mass production. Sadly, we haven't seen the board in person yet, and we’ll add better photos of this beast later once we see it at their booth.

The next motherboard they showed off was the X870E Aero X3D Dark Wood. As the name suggests, it’s an Aero X3D Wood, but with a darker-colored synthetic wood material for an elegant look and a real wood feel. Gigabyte also expanded the Stealth series (back-connector) motherboards with the world’s first gaming B850 MicroATX boards: the B850M Aorus Stealth and the Stealth Ice (white). Shrinking the boards down lets you get that stealthy, clean look in a smaller chassis, and if we’re lucky, they should cost less since they are based on the B850 chipset and are smaller. ALBM - Infinity GPUs

Last but not least, the Infinity cooling architecture found on the flagship Aorus RTX 5090 Infinity video card expands down the product stack. It is now available in RTX 5080, 5070 Ti, and 5070 flavors, keeping the same Windforce Hyperburst cooling design (said to deliver much more airflow than standard hollow-out and non-hollow-out cards), a hidden power connector, and a unique appearance compared to most that are simply cuboid. If you’re a fan of the wood look on the Aero boards, Gigabyte has you covered, also showing off the Aorus RTX 5080 Infinity Wood that takes the Infinity design and adds walnut ‘wood’ accents to match motherboards.

Gigabyte had a lot to show off at Computex 2026. While there was a new AI ecosystem, most of it was for consumers like us, which is nice to see compared to the AI-laden showcases at recent trade shows.

During a press Q&A held at Computex 2026 this morning, Nvidia CEO Jensen Huang hosted a wide-ranging discussion of why the company is entering the PC market with its just-announced RTX Spark platform, and what it stands to gain by introducing an all-new chip into the already crowded personal computing space.

In a response to a question posed by analyst Ryan Shrout, Huang said that the decision to develop and introduce a new PC platform with RTX Spark isn't fundamentally about business concerns like the potential margins involved, and that "we don't really have to choose between solving one problem or another."

Huang repeatedly emphasized that computing at every scale, from the PC to the data center, is undergoing a fundamental shift from a world where systems sit and wait for us to use them to an agentic loop where they'll autonomously work to complete tasks for us by running AI agents and models that can call tools and use Windows and applications themselves. Vera Rubin is the architecture for that shift at data center scale, and RTX Spark is the engine for powering that loop for the PC.

Huang envisions an RTX Spark-powered future where he'll just talk to agents running on his PC via WhatsApp, and they'll get things done for him and communicate the results of that work back to him. "Tell me that's not R2-D2. Tell me that's not robotics. Tell me that's not cool."

He says the company is driving this shift because it has "a chance to reinvent the single most important instrument, the single most important tool of humanity" with RTX Spark PCs, and "we're not going to sit around and not let it get done." He further elaborated that Nvidia sees the opportunity to make a significant contribution to personal computing's future with RTX Spark, to solve a hard problem, and to do it "insanely well." The ultimate question, as Huang sees it, is "Can we create something the world would love?"

Although the highly integrated CPU and GPU and unified memory architecture of RTX Spark might look broadly similar to Apple Silicon, Huang dismissed the idea that the company is trying to compete with Apple and products powered by its M-series chips. He says that Apple has a "world-class silicon roadmap," and that it's building those chips in service of the needs of its own unique device, hardware, OS, and application ecosystem. He says that Nvidia's goal is to "reinvent the PC," and that its focus is "100% on Windows."

Huang also tried to assuage concerns about Nvidia's long-term commitment to the RTX Spark platform, given the relative lack of purchase that other Windows on Arm devices have achieved in the market thus far.

Huang said that "once we start a new product line, once we start a new software image, we support it for as long as we shall live." He cited the long-lived Nvidia Shield TV platform as an example of how the company "takes great care" of the software of its devices, and he says that will be true for RTX Spark devices, as well.

He asserts that the software stack for RTX Spark "is likely the best software stack provided ever, and the software stack defines the experience of the user these days." He says those stacks are the reason why GeForce, Quadro, and RTX Pro products are already "deeply loved," because "we take care of the software."

Huang also touted the breadth of hardware companies that have signed up to make Spark systems as a vote of confidence in the future of the platform. Referring to the laptops on stage with him from Asus, Lenovo, Dell, MSI, Microsoft, and HP, he boasted that "this is literally every computer maker in the world," and "we have never seen anything like it. No new product, no new chip has ever been launched where this much of the world's computer ecosystem signed up."

Beyond the laptops and desktops Nvidia and its partners have already announced, Huang also notes that the RTX Superchip is the SoC formerly known as N1X, and that it has a second, smaller chip called N1 yet to be detailed. He also described N2 and N3 Spark chips for future systems that will power future AI PCs, a commitment he first revealed during his Monday Computex keynote.

Huang said, "We're going to expand our family... We're going to expand the footprint of this architecture, and we're going to extend this architecture for a very long time."

If you take a look at Tom's Hardware this week, you might recognize that we've been posting many reports directly from the ground at Computex 2026 in Taipei, Taiwan. This year, we're offering readers a chance to take a look at how we're coping amid the sweltering Taipei heat and telling you exactly what we've been up to each day.

While new announcements (and a lot of writing) are part of the overall Computex experience, we hope you enjoy this series of blog posts from our team on the ground. If you haven't caught up on our Day 0 post, be sure to check it out first. With the busier trade show days looming, Tom's Hardware's fearsome five marches on into the depths of the Nanggang Exhibition Center and beyond.

Paul Alcorn: Editor-in-Chief

The pace of the show is quickening as we get closer to the first day of the show floor opening, which occurs tomorrow. Today found me scrambling after the Nvidia keynote to make it to the Qualcomm keynote, and then the following press question and answer session. Qualcomm really didn’t bring many new announcements to the show, so there weren’t any terribly big news gems to be found. My marathon demo run through Nvidia’s suite, which took two hours, was a lot more interesting as the company demoed a seemingly unending string of gaming and AI demos, with most of those powered by the new Microsoft Surface powered by RTX Spark. In the end, I ended up back at the hotel at 10 pm for some rest before a 7:30 am question and answer session with CEO Jensen Huang tomorrow.

Matt Safford: Managing Editor

I spent the early morning writing, before grabbing Mos Burger for breakfast and heading to the Nangang Exhibition Center (Computex HQ) to pick up my badge and take more photos for daily wrapups and our Best Of story. After showing our Computex rookie Joe Shields around a bit, we headed to Gigabyte's Computex kickoff, where the company showed off many things, including, most notably, the X870E Aorus Infinity Next, a motherboard wrapped in 3D-printed metal.

We were told that just the production and materials of this board cost thousands of dollars, so it won't be making it into your next build (unless maybe you're a billionaire), but it is interesting to see Gigabyte push the boundaries of what is possible.

Joe Shields: Staff Writer, Components

I have to talk about the night market I went to last night first. There was so much to see and even more to eat! So many different foods and vendors selling wares, carnival-style games, and there were a lot of people. I ate a pork pepper bun, fried prawns (would recommend both), and what was supposed to be a brisket burger that I think ended up being chicken sausage. Getting there (and around in general) on the subway was a lot easier than I expected, as most signs/announcements had an English translation and were color-coded.

Today was a day of unrest before the storm, with two events in the late afternoon, including one with Gigabyte, who showed off a lot of cool goodies, including the X870E Infinity Next with its metal 3D printed heatsinks — the star of the show for me so far. At this point, I’m sleepy, as the dramatic 12-hour time change has finally caught up to me. Tomorrow, the fun really begins, starting with a visit to Cooler Master HQ and then to the convention center and booth hopping for the rest of the day.

Jake Roach: Senior Analyst, CPUs

Day one of Computex is down, and it was a busy one (a sign of things to come). I started my day chatting with Intel’s Nish Neelalojanan about a broad range of consumer topics, and we’ll be rolling out some choice quotes from that interview over the coming days (the first is already live). After some time at the Qualcomm keynote, I went back over to Intel to talk Xeon 6+ and the company’s new Arc G3 Extreme.

It’s hard to overstate just how convenient Taipei is when it comes to darting around the city like this. The MRT (train) is fast, cheap, and always on time, allowing me to get around to various places while (mostly) avoiding the Taiwan heat and endless Uber bills.

Today was Intel, tomorrow is AMD. We have a series of roundtable interviews set up, so hopefully we’ll get some more interesting insights into where the x86 gang stands, especially in the face of the RTX Spark announcement from Nvidia.

Jeffrey Kampman: Senior Analyst, Graphics

Today was all about Nvidia for me, from early morning until late at night. In the morning, I joined the throngs of GTC Taipei conference-goers to hear CEO Jensen Huang talk about the company’s continuing full-court press for the data center with Vera Rubin.

We also finally learned all the juicy details of the RTX Spark platform (aka N1) and the company’s considerable efforts to spark a Windows on Arm revolution across hardware, operating system, and software. Even if you’re skeptical about Nvidia’s agentic AI vision for the future of personal computing, it’s impressive that it’s gotten everybody who needs to be on board for a seamless Windows on Arm experience on board, and the platform already feels quite mature. We’ll be spending more time with Nvidia tomorrow between a Jensen Huang Q&A and more hands-on opportunities with RTX Spark.

AMD finally reintroduced the Ryzen 7 5800X3D at Computex 2026, more than four years after it originally launched in a bid to combat rising DDR5 prices. Despite the re-release looking like a simple product spin-up, AMD’s David McAfee, VP and general manager of Radeon and Ryzen, says that “a whole body of engineering work” went into the re-release, as the original bonding process TSMC used for the Ryzen 7 5800X3D was no longer available.

“It's not as simple as just bringing back the 5800X3D,” McAfee said. “The original stacking process that was used at TSMC changed when we went from first-gen to second-gen cache, so we had to re-engineer that product, and there actually was a fair amount of development that went into bringing back the 5800X3D.”

The Ryzen 7 5800X3D used TSMC’s SoIC or System-on-Integrated-Chips hybrid bonding technology. It uses a combination of “hot” and “cold” bonding to marry two pieces of silicon together, which then share power with through-silicon vias (TSV). Fundamentally, this connection hasn’t changed over the course of 3D V-Cache’s existence, but it has evolved. With the move to Ryzen 7000, AMD had to make some changes to the 3D V-Cache design, which it then carried forward with Ryzen 9000.

To avoid any confusion, when we talk about second-gen 3D V-Cache here, we’re not talking about AMD’s new packaging available on Zen 5 CPUs, where SRAM is placed under the CCD rather than on top, as is the case with Zen 4 and 3 X3D chips. We’re talking about the bonding process that AMD used at TSMC, which changed from first-gen X3D chips to second-gen X3D chips.

“It completely changed the characteristics of how those two pieces of silicon are bonded together and how they were stacked together, and so when that first-gen facility really kind of went offline, then it meant there was a whole, you know, body of engineering work that had to be done to understand if we could even migrate the 5800X3D to the new, second-generation stacking process,” McAfee said.

It’s possible that AMD intended to bring back the Ryzen 7 5800X3D sooner, though McAfee stopped short of saying that outright. The shift in packaging helps explain the Ryzen 7 5800X3D’s (and eventually the 5700X3D’s) absence from the market. The chip has seen spotty availability over the past two years, and it’s been completely sold out over the past year, with resellers demanding as much as $800 on the secondhand market.

“That's been a lot of the work that's kind of been going on in the background to get us to where we are today, is redoing the qualification of that stacking process, building samples, testing to make sure that the reliability is top notch for consumers who might want to buy this product, and then, you know, kind of rolling it out and ramping it into production again in a new process of stacking those dies together,” McAfee told Tom’s Hardware.

You can read more about second-gen SoIC in our SoIC roadmap on Tom’s Hardware Premium, though it has far more implications for the data center (at least currently) than for consumer chips. Regardless, AMD couldn’t simply reintroduce the Ryzen 7 5800X3D; instead reworking the chip to work with TSMC’s second-generation stacking process. McAfee says it ended up being “a labor of love” for the engineers to work on this part again, as the company went through testing and validation for a re-release.

Intel’s Arc G3 chips are gunning for the AMD-dominated, high-tier integrated graphics market that has become such an important enabler of the modern handheld PC gaming experience. But as high-memory prices push up the costs of even entry-level discrete GPUs, there could be much more of a place for powerful onboard graphics in the PC gaming landscape in the years to come.

We sat down with Intel’s Senior Director of Product Management, Nish Neelalojanan, in Taipei, Taiwan, at Computex 2026 to talk more about the G3’s development and how it fits into Intel’s lineup. Here, we're presenting the full transcript of our conversation.

This transcript has been lightly edited for clarity.

Jake Roach, Tom's Hardware: So what was the idea behind the G3, because you guys have tried before, right? I believe it was with MSI? And now you’re putting a bigger emphasis behind it with a whole new branding.

Nish Neelalojanan, Director of Product Management, Intel: It was a combination of two things. So first of all, we started trying already with Meteor Lake, and yes, we were experimenting. This was all standard off-the-shelf parts, and we learned a lot as we came into Lunar Lake. The power management for handheld needed to be more customized, so we started tweaking further, and as we got into Panther Lake, the architecture lent itself to lower power gaming. We moved the E-Cores onto the performance cluster, so you have E-Cores both on your efficiency island and your performance cluster, that means your E-Cores have access to L3 cache, so E-Cores are now performant enough to run games.

A lot of the time, in a low-power scenario, you are more GPU-bound than CPU-bound because the GPU is power starved, so if you can reduce the power on CPU and dump it on the GPU, you'll get much better performance. So, with that architecture change with Panther Lake, now is the perfect time. All the goodness we've learned, we can capitalize on it. We have a silicon architecture, we can lend itself to low power, and we have big enough graphics now…

Jake Roach: A really impressive iGPU.

Nish Neelalojanan: So, that is what had the impetus on, hey, what if we did a CPU line, which is graphics first, or leading with a very big graphics, but small enough CPU that doesn't grab enough power, but good enough to run all your handheld games. It's great for handheld gaming or non-PC form factor, running low-power gaming. So we wanted to start a line of products, which would be integrated graphics forward, with the right CPU.

Jake Roach: And these are wholly unique entries, right? If I remember correctly, there's no 14-core Panther Lake.

Nish Neelalojanan: These are completely unique chips. So they are based off the same die, but we've optimized it with, like I said, core count, so that taking two P-Cores off, because most of the games are going to run on the E-Cores on the performance cluster, you also cut down on different I/Os, so you don't need as many ports on a handheld as you would need on a laptop, right, so you cut down, so it's cutting down all the things you don't need.

Jake Roach: Really focusing it on that form factor.

Nish Neelalojanan: Yeah, that will be on the hardware, and then software-wise, we have a lot of other software optimization. So, now in order to have them pinned onto the E-Core, we have a BIOS control optimizer, so extra ways to have your thread director direct your game threads onto your E-Cores.

It's basically making sure we are directing the game threads onto the E-Core. [We also have the] ability to do power gating, so that we have features like endurance gaming, which we had on the laptops. Now, for handheld, we've added some features, so you can go with different presets. You can say, I want 60 frames per second, and then it will optimize your profile accordingly, or I want 30 frames per second. So you have a frame cap, and then your SOC resourcing is optimized, so that you will increase your battery life 2, 3, 4, hours.

Jake Roach: Battery life is so important for a handheld, right? I was playing a little Forza Horizon 6 on the plane coming over, and one way that I'm doing that right now on Linux is with Lossless Scaling, with frame generation in any game. As you're saying, apply that 30 fps cap frame generation to the mix, and you can get really good perceived performance.

Right now, you guys have multi-frame generation through XeSS 3 through specific games, but there's no driver. Is that something you're looking into, given how important that can be for the local gaming experience?

Nish Neelalojanan: So, 100 plus games have already enabled MFG, but you could imagine, as you said, it's important. So we're exploring, but as we get closer, we'll talk more about when and where it intercepts.

Jake Roach: The other thing I did want to ask about was form factors, which is something you kind of hinted at. Right now, with component prices being so high on just a typical DIY PC, we're seeing a really big push for budget laptops for people that maybe don't need as big of graphics and handhelds for people that really care about gaming as a laptop replacement. I'm just curious, kind of broadly, what you think about the dynamics between these form factors? Is this something that is just a temporary market connection, given that prices are so expensive, and are you planning around that, or is it something more long-term? Where do you think these are going to be the preferred form factors?

Nish Neelalojanan: I think, from a budget-conscious perspective of value buyers, our core 300 series, I think probably this is the first time in a long time the mainstream is getting some of the new ideas, right? So, today if you take a value-conscious segment in the past, it was always, hey, you have the big innovation, which we launched, it gets waterfall down. But as the innovation started getting expensive more and more, that waterfall did not happen, it was basically take the old chip, do some minor updates. So we wanted to take all these new like battery life performance uplifts, and you know, having the new AI updates, all of that, but to be able to be affordable, that's what is Wildcat Lake, or Core 300 series. So that's kind of for the budget-conscious buyer, we wanted to make sure we put some new IPs out there, because I don't think anyone is putting that out.

So that's part A. Part B of your question is handheld as a form factor. I think handheld as a form factor is interesting. Different people are trying to do multipurpose use, so would it ever go from companion to main? TBD. But can it expand its use cases from, hey, can I have a handheld, can I have a docked experience? I think long-term, yes. Currently, the software interfaces and a lot of the, let's say, ecosystem around it needs to evolve for it to be meaningful, but there's a lot of experimentation around dock experiences and stuff, which we are working with partners to experiment, but as it stands, I think handheld alone as it's gaming first.

A lot of our partners are experimenting; they're having all those capabilities available. How can you dock, how can you connect keyboard and mouse directly, and then be able to do it, because, like you said, costs are going up. If someone buys this, they want to be maximizing it.

Arrow Lake refresh's positioning

Jake Roach You touched on Wildcat Lake in the mobile segment, and recently we had Arrow Lake refresh, which was a really big readjustment in pricing. In particular, I'm just kind of curious to get more color on that, because when I reviewed those chips, I the expectations, I was briefed on them, and I was like, I know what to expect, but it was a very different Intel than I was used to working with on the desktop run.

Nish Neelalojanan: Short answer would be, we wanted to make sure we are putting out things which gamers would care about and show that we care about gamers, so this was an attempt or step one in getting to that expectation, right? And then obviously same thing with handhelds, making sure we are putting something which the gamers would want, so that's the highest level.

Jake Roach: I mean, again, seeing the results, we've essentially made the original Arrow Lake line outside of a few chips, irrelevant with these two chips. The 270K Plus scales above 285k and 250K can go toe to toe with either the Core Seven or Five.

Nish Neelalojanan: That’s a good thing, right?

Jake Roach: That is a good thing, but it's a good thing for us. It is not a good thing for Intel, right? Like, internally, that undermines your own product. And so I'm curious about the decision there, because at some point you had to have had pushback on, hey, this $300 Core 7 is going to undermine our $600 Core 9 flagship.

Nish Neelalojanan: It was just a decision made with end users in mind, and we want to make sure we are providing value as we come out, and if at all we need to start somewhere, right? And in terms of desktop, that was an effort to let's go with value focus first, and that will help us then gain confidence. From an enthusiast perspective, we needed to build back our reputation. I am sure you would agree with that, and this was, we’re making sure we are providing value to the gamers, and we start with Arrow Lake Refresh, and we have a very strong roadmap to come, so we want to continue.

Jake Roach: It did seem like an appetizer, almost, given you know everything that happened with Arrow Lake, and yeah, I appreciate you wearing that a little bit, because those were again, they were very interesting parts for a number of reasons.

Nish Neelalojanan: Savior [Kim, Intel Director of Client Communications] can correct me if I said something I shouldn't, but that was kind of highest level.

Leaving Hyper-threading behind

Jake Roach: Another question I had; This is more on the mobile side of things, or SoC side of things, is around hyper threading. So I have this question for the Xeon folks that I'm meeting with later today, because there have been comments in the financial reports, comments about returning hyper threading to the data center, whole lot of stuff around that on the consumer side of things. You guys left hyper threading behind with Arrow Lake. I'm wondering now that we have a desktop generation, a mobile generation under your belt. What do you see with that move to get rid of hyper-threading? And is there any consideration for maybe going back to some form of SMT in the future?

Nish Neelalojanan: Highest level, our decisions are always are we getting the right level of performance. The best way to achieve that performance is what we want to go with. Like you said before, with Arrow Lake Refresh, you're not only getting the right game performance at the right price point, but you're also getting almost 2x multi-threaded performance compared to competition, right? So, if you can deliver that without SMT, though the end user, it doesn't matter to the end user. In fact, you're actually getting even better multi-threaded performance because they're actual physical codes versus virtual threads, right? So that's where I would leave it at. We always reevaluate, but it's the best way to give that level of performance in that given price band or that given SKU. So we continue to keep re-evaluating, and different segments may need different things.

Jake Roach: Right now it sounds like it's working out well?

Nish Neelalojanan: Yeah, and like in terms of all the different agentic AI workloads, you need CPU as an orchestrator, having nth number of threads, cleaning up data, lining up a memory, a lot of threads help. So, like I said, when there is utility, and when there is a need, we will constantly evaluate it's, it's rigid to say, oh, it's behind us, or it's rigid to say, oh, we are going to run towards it: If it makes sense, it makes sense, yes. That’s where the data center decision is. They talk more about the growing workload, there is a need.

Jake Roach: It has been interesting. We're coming up at Tom's Hardware on 30 years, and we did a retrospective on CPUs, so I went back to the very first Pentium Two review on Tom's Hardware, and seeing the hyper-threading, and how it was used over decades, it was really fascinating to look at.

Nish Neelalojanan: A lot of the low-power segments, like now, handheld, yes? Those eight E-Cores on that performance cluster are significantly helping with all the low-power gaming, right? So, a lot of these decisions are paying off as it stands. As the workload evolves, as we evolve into different architectures, we will have to evaluate based on at that time what would be the right decision. Okay.

Jake Roach: For years now, Intel Foundry has laid out a really aggressive foundry roadmap. We saw 18A first and now we finally have 18A in data center with Xeon Six Plus. Is that the kind of the cadence we should expect going forward for Intel's cutting-edge hooks to see them debut first on the consumer front?

Nish Neelalojanan: It's the same answer I said before wherever it makes sense first. So we've got especially a lot of our consumer client CPUs, we pick the right process node, which made sense for the right tile, especially now we have multi-chip solution. It gives us the flexibility to pick and choose the right process node for cost, readiness, optimizing for R&D, because sometimes you won't have that IP on a different process, so it's easier to just reuse it, based on availability. Sometimes it's costing some bigger tiles, you can put it on latest and get it to get performance. Some of the tiles where you don't need to push frequency as much, you put it on an older node. And now with all the supply in and around the industry, picking the right process nodes, which is more available, is also going to be important. So we always go through all of these considerations and pick and choose, so there is no settling on client will start data center will follow, vice versa. It's we pick the right process choice based on that architecture, for that side.

Reacting to Nvidia's RTX Spark

Jake Roach: I want to get your reaction to [Nvidia’s RTX Spark]. If you need any better reminder that Twitter is not real life, there's a lot of talk on Twitter that Nvidia entering this market completely decimates and it rules everything. I don't think that's true, but I want to see your reaction to Nvidia getting into that space.

Nish Neelalojanan: I mean, Nvidia puts out great products, and they know how to do gaming. They know how to do all these different things. So we always take everything with a healthy dose of paranoia, but we are also very, very confident with our products, in the sense that X86… Let me put it this way, when we entered this discrete graphics business, our graphics business, it took a painful few years for us to work through all the drivers, all the compatibility issues, and everything ironed out, same thing goes on when an ARM CPU enters a market that's going to be tons of compatibility DRM issues, backward compatibility As a result, we are very confident that we have the right CPU, GPU mix for clients, both for gaming and when it comes to what you call AI inference workloads.

That said, Nvidia is a great partner. We will continue to work with them. You saw some of our announcements. We have some longer-term commitments with them, so both of us have different parts of the roadmap that we will expand together, where there'll be a roadmap where we will be partnering, and where there might be places where we will be competing, but I think it's great for the industry that there is different choices.

Jake Roach: I know, it's a weird situation, especially for Intel, because you guys, you guys do work with Nvidia. Yeah, when I pose similar questions to the other guys, they, they're a little bit more fiery in their responses.

Nish Neelalojanan: Compatibility is going to be a key thing there. x86 on the CPU side is going to have a lot of advantages. We talked about some of the new instruction sets, which got announced by the x86 Consortium, a lot of those lend itself as much to gaming as much as AI, and you'll see a lot of that being talked about more.

A lot of it were agentic AI examples and stuff, because you have to say AI three times before you can talk about anything else, but they also help with gaming significantly, so yeah, it doubles the amount of registers, which you would execute one instruction, so it's based off of AVX, but there’s a few others which came out with it.

The health of the consumer PC market

Jake Roach: Finally I want your reaction more broadly to the PC market right now, because we have all the rising component prices, we have very expensive laptops. On the desktop, it's really, really hard to build a PC right now. I think motherboard sales are down some 30-40% I know you're releasing products to address that market between Wildcat Lake and Arrow Lake pretty refresh, but I kind of want to see your reaction to how that pans out over the next maybe three to five years. Is it a continual area of focus, or is it something that hopefully we're just dealing with over the next few years, where we're really focusing on the budget segment?

Nish Neelalojanan: Large memory is completely overshadowing any CPU prices, right? Memory and storage. The CPU is not anymore determining your system price point, and when you're paying that amount, people will obviously start upgrading. Now, that said, there are still Panther Lake systems you can get below $1,500 out there, right? It's going to be dependent on OEM. It's going to be dependent on markets, and even the Wildcat Lake, they'll announce a $599 starting price point. Yeah, so there are definitely designs which are coming at comparatively reasonable price points, which are available, and longer term, I think something has to give right. The over inflation, we will have to keep an eye, but if I could predict the memory market, I would be rich in stock!

Jake Roach: Let me phrase the question a little bit better, because are you making plans for a longer term, a longer term squeeze on the consumer front, because surely you're going to have to make those plans if you see the headwinds going that way.

Nish Neelalojanan: We do have products with support for DDR4 both on desktop and mobile, so Raptor Lake, you're not end of life in any of them, they're there. We'll continue to make sure that there are products which can take care of older memory technologies if they're available and cheap. Second thing is, we are making sure we are validating lower configs as well. Wildcat Lake starts at 8GB, Wildcat Lake is a single channel product, so there are products which can leverage with low memory and give reasonably good performance, so we are doing everything we can from our perspective to be able to help in any small way. But like I said, when CPU becomes the least relevant from an overall BOM (Bill of Materials) perspective, because it's so expensive. Then we also have CPUs you can buy out into that.

Jake Roach: Speaking of memory, I don't know if you had any involvement with half ranked? Is that what they call them, half ranked DIMMs from ASRock? It was with ASRock and Intel.

Nish Neelalojanan: I am not familiar with that, but we are working with a lot of indigenous memory suppliers to validate them, so we’re doing everything we can in terms of it's not just one, two, or three. If there are some local specific memory vendors, but like in PRC, and now Indonesia is even bringing up a couple of them. We're trying to validate as much as we can, so there's enough choice that people can get pockets of relief. Right? We are looking at UFS for a longer-term horizon, so that every little thing helps, right?

Jake Roach: Absolutely. I appreciate you taking the time and talking over everything. I'm very excited to see the G3 chips in action. I saw them yesterday at the Acer showcase, and I played a little bit of Forza Horizon 6, and these are pretty good.

Nish Neelalojanan: With G3 at least we're putting out some latest and greatest stuff, and in terms of a lot of these, it's not necessarily exclusive. We're broadly available, the 12 Xe on the PC side, and on the handheld, it's not like limited to one OEM. Unlike some people who hold it back, just only give it to one OEM.

Jake Roach: So all these handhelds, I believe all the ones that announced are all Windows-based handhelds. Is there consideration for Linux? How much consideration or weight do you put on that, given things like Steam OS proper?

Nish Neelalojanan: So highest level stuff we announced now is Windows based, but you can take those devices and install… and I'm sure you would imagine we would continuously want to make sure that those experiences are reasonable for end users, and we are, we would talk more about as we get closer to something, but we are exploring beyond Windows, and as we get closer, we'll talk more about.

[Session ends]